Norwegian tech company AutoStore today unveiled the latest version of its warehouse storage and retrieval robot, as it seeks to boost the efficiency of its automated fulfilment system.
Dubbed the P5 Pro, the electric cube-shaped robot is fitted with next-gen lithium titanium oxide batteries. The company says this allows the little robotic workers to toil harder and longer. The improved battery technology also enables AutoStore’s customers to cut the number of charging points in their warehouses by 85%, boosting storage capacity.
“Retailers are constantly looking for new ways to adapt to rising customer demand for more, faster delivery of goods. These more efficient robots can help them fulfil this need,” Carlos Fernandez, the company’s chief product officer, told TNW.
Chief product officer Carlos Fernandez stands next to the new P5 Pro robot. Credit: AutoStore
Founded in 1996, AutoStore has developed a cube-based storage system where small robots navigate a grid-like structure to retrieve and store items in bins. Unlike a normal warehouse where items are stacked on shelves and workers have to climb up a ladder or use a forklift to retrieve each product, in an AutoStore the products are delivered straight to a station where a (human) worker packs them ready for shipping.
The <3 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
Letting robots do the legwork isn’t just faster and safer, it also means the warehouse can be much, much smaller. The system uses four times less space than a manual setup, and increases productivity by up to five times due to time savings, said the company. It also frees up employees to take on less physically demanding jobs.
“There’s a misconception that automation is bad for jobs, but I think it’s quite the opposite,” said Fernandez. “Our robots don’t take people’s jobs, they allow employees to do better, safer ones.”
Currently, some 45,000 AutoStore robots are operating in 1,200 sites across 50 countries serving over 800 customers — including the likes of IKEA, DHL, Decathlon, Gucci, and Puma.
In 2019, AutoStore was sold to American private equity firm THL Partners for NOK 16bn ($1.88bn), becoming Norway’s first unicorn company in the process. A couple of years later, the Softbank-backed firm went public at a value of $12.4bn, making it one of Norway’s largest-ever IPOs.
For Fernandez, key to the company’s success is flexibility. “When customers are thinking about automation they need a solution that is flexible enough to grow and change with their business,” he told TNW.
The modular, LEGO-brick-like design of the AutoStore system means that it can fit to any sized warehouse, and can be expanded quickly, to accommodate for, for instance, a surge in demand during the Christmas holiday period.
As coronavirus restrictions led to a surge in online shopping, demand for the company’s products shot up. Sales jumped 88% in the first half of 2021 to about $150mn, AutoStore told Bloomberg.
And now, amid an economic downturn, warehouse operators are looking for ways to cut costs, which is exactly what AutoStores’ latest robot was designed to do.
The P5 Pro has already been trialled in the warehouse of existing customer Boozt, a Swedish fashion giant which operates an existing fleet of over a thousand AutoStore robots at its facility in Malmö.
Orders for the new robot will open up from next month, with first deliveries beginning at the end of 2024, said Fernandez.
There’s a looming global computing capacity crunch that cannot be sustainably addressed the way we’re doing things right now.
Simply put, between artificial intelligence (AI) models growing exponentially and an ongoing global digital transformation, data centres are running out of space. Their vacancy rates are hitting record-lows and prices are rising in response to demand, which is cause for much unease among tech leaders.
If this trend continues, at some point, we will reach a juncture where we can no longer accomplish all the things that technology theoretically allows us to do, because our capacity to process data will be constrained.
Perhaps the biggest worry is that AI’s transformative potential, which we’re only just beginning to tap into, will be throttled by purely physical constraints. This will hinder new discoveries and the development of more advanced machine learning (ML) models, which is bad news for all, except AI apocalypse alarmists.
Is there any way to avoid the computing capacity crisis? Since massively scaling back our computational demands isn’t really an option, the only alternative is to significantly boost capacity, which boils down to two available courses of action: build more data centres and develop better digital infrastructure.
But that’s easier said than done — here’s why.
Why more data centres isn’t the answer
Until now, increasing demand for computing capacity has been, in part, met by building more data centres, with conservative estimates putting real estate taken up by data centres growing at ~40% per year. It’s a figure that you can expect to remain fairly steady, as supply issues, power challenges, and construction delays are severely limiting scaling capacity expansion.
In other words, today, demand cannot be simply met by ramping up data centre construction.
Nor should that be something we aspire to. Each of these football-field-sized warehouses gobbles up gargantuan amounts of energy and water, placing severe strain on the environment, both locally and globally. A single data centre can consume as much electricity and water as 50,000 homes and the cloud’s carbon footprint already exceeds that of the aviation industry.
Credit where credit is due — data centres have come a long way in minimising their environmental impact. This is in large part thanks to a fierce sustainability race, which has propelled innovation, particularly as it relates to cooling and energy efficiency. Nowadays, you’ll find data centres in underground mines, in the sea, and using other natural cooling opportunities such as fjord water flows, all to reduce energy and water consumption.
The trouble is, this isn’t scalable globally, nor is boiling our seas a viable path forward. Erecting more data centres — no matter how efficient — will continue to wreak havoc on local ecosystems and impede national and international sustainability efforts. All while still failing to meet the demand for compute resources.
Still, two chips are better than one, unless…
Think inside the box
… unless that single chip operates at twice the speed. To avoid the capacity crunch, all hopes rest on improving the digital infrastructure, namely, the chips, the switches, the wires, and other components that can improve data speeds and bandwidth while consuming less energy.
Let me reiterate — the evolution of AI depends on finding ways to transfer more data, without using more energy.
Essentially, this means two things. First, the development of more powerful and AI-centric chips. Second, the enhancement of data transfer speeds.
1.Designing custom chips for AI
Existing digital infrastructure isn’t particularly well suited for the efficient development of ML models. General-purpose central processing units (CPUs), which continue to be the primary computing components in data centres, struggle with AI-specific tasks due to their lack of specialisation and computational efficiency.
When it comes to AI, graphics processing units (GPUs) fare much better thanks to better processing power, higher energy efficiency, and parallelism. That’s why everyone’s snatching them up, which has led to a chip shortage.
Yet GPUs inevitably hit the same brick wall. They’re not inherently optimised for AI tasks, leading to energy waste and suboptimal performance in handling the increasingly intricate and data-intensive demands of modern AI applications.
That’s why companies such as IBM are designing chips tailored to AI’s computational demands that promise to squeeze out the most performance while minimising energy consumption and space.
2. Improving data transfer capacity
No modern AI model operates on a single chip. Instead, to get the most of available resources, you assemble multiple chips into clusters. These clusters often form a part of larger networks, each designed for specific tasks.
Accordingly, the interconnect, or the system facilitating communication between chips, clusters, and networks, becomes a critical component. Unless it can keep up with the speed of the rest of the system, it risks being a bottleneck that hinders performance.
The challenges for data transfer devices mirror those for chips: they must operate at high speeds, consume minimal energy, and occupy as little physical space as possible. With traditional electrical interconnects fast reaching their limits in terms of bandwidth and energy efficiency, all eyes are on optical computing — and silicon photonics, in particular.
Unlike electrical systems, optical systems use light to transmit information, providing key advantages in the areas that matter — photonic signals can travel at the speed of light and carry a higher density of data. Plus, optical systems consume less power and photonic components can be much smaller than their electrical counterparts, allowing for more compact chip designs.
The operative words here are “can be.”
The growing pains of cutting-edge tech
Optical computing, while extremely fast and energy-efficient, currently faces challenges in miniaturisation, compatibility, and cost.
Optical switches and other components can be bulkier and more complex than their electronic counterparts, leading to challenges in achieving the same level of miniaturisation. As of now, we are yet to find materials that can act as both an effective optical medium and are scalable for high-density computing applications.
Adoption would also be an uphill battle. Data centres are generally optimised for electronic, not photonic, processing, and integrating optical components with existing electronic architectures poses a major challenge.
Plus, just like any cutting edge technology, optical computing has yet to prove itself in the field. There is a critical lack of research into the long-term reliability of optical components, particularly under the high-load, high-stress conditions typical of data centre environments.
And to top it all off — the specialised materials required in optical components are expensive, making widespread adoption potentially cost-prohibitive, especially for smaller data centres or those with tight budget constraints.
So, are we moving fast enough to avoid the crunch?
Probably not. Definitely not to stop building data centres in the short term.
If it’s any consolation, know that scientists and engineers are very aware of the problem and working hard to find solutions that won’t destroy the planet by constantly pushing the boundaries and making significant advances in data centre optimisation, chip design, and all facets of optical computing.
My team alone has broken three world records in symbol rate for data centre interconnects using intensity modulation and direct detection approach.
But there are serious challenges, and it’s essential to address them head-on for modern technologies to realise their full potential.
Professor Oskars Ozoliņš received his Dr.sc.ing. degree in optical communications from Riga Technical University (Latvia) in 2013 and received a habilitation degree in physics with a specialization in optical communication from KTH Royal Institute of Technology in 2021. He is the author of around 270 international journal publications, conference contributions, invited talks/tutorials/keynotes/lectures, patents, and book chapters. You can follow him on LinkedIn here.
A new AI tool that predicts viral mutations could guide treatments for COVID-19 — and the next pandemic.
The system, named EVEscape, was developed at Harvard Medical School and Oxford University. In tests, the tool accurately predicted the most concerning and frequent variants of the SARS-CoV-2 virus that emerged during the pandemic.
A study published last week in Nature revealed an array of promising results. EVEscape’s forecasts proved more accurate than experimental approaches, while faster and more efficient than lab-based tests. The tool also successfully pinpointed therapies that would struggle to subdue new variants.
The predictions are already informing pandemic monitoring efforts. For over a year, the researchers have been releasing biweekly rankings of the most concerning new SARS-CoV-2 strains. The findings are shared with groups including the World Health Organization (WHO).
“There are still thousands of new strains emerging each month — too many to experimentally test,” Pascal Notin, an Oxford University researcher who co-authored the study, told TNW. “EVEscape allows us to rapidly determine the threat level of the new strains.”
Notin and his colleagues have also used EVEscape to successfully predict mutations of HIV and influenza. They’re now testing the tool on lesser-known that could also cause pandemics, such as Nipah and Lassa.
In the future, the researchers envision EVEscape informing vaccine design. At present, vaccines and therapeutics are tested retrospectively against previous pandemic mutations.
EVEscape could add evaluations on where the virus might go next. This offers hope for a powerful new treatment: variant-proof vaccines.
How EVEscape predicts virus mutations
The new tool is based on a generative model called EVE (Evolutionary model of Variant Effect).
Initially, EVE was developed to predict the risks of genetic mutations causing human diseases, such as cancers. When COVID-19 proved alarmingly adept at mutating beyond the constraints of treatments, the researchers adapted their model to SARS-CoV-2.
Generative models have unique strengths for this job. A key aspect of predicting which mutations will evade immunity is whether they will preserve the so-called “fitness” of the viral protein. This fitness leads to a functional protein that expresses, folds, and binds to the host cell receptor.
“Generative models trained on evolutionary sequences are critically helpful in supporting that prediction,” said Nodin.
To make the predictions, EVE is trained to learn a compact representation of certain protein sequences. These sequences have maintained their fitness over thousands to millions of years of evolution.
“By doing so, it has to implicitly learn the biochemical constraints that underpin functional sequences,” Nodin explained. “We can subsequently leverage that understanding to predict whether new mutated forms for the protein — whether they are human or viral proteins — will be functional themselves.”
AI’s future fight against viruses
EVEscape’s adaptability stems from its simplicity. The tool learns from a dataset comprised of only viral protein sequences and their structure. As a result, it can be applied to any virus and at the very start of a pandemic.
“This is a great improvement over methods that typically need to wait for relevant pandemic antibodies to be broadly available for testing,” Nodin said.
Another string to EVEscape’s bow is its modular design. When more powerful generative models emerge, the current fitness prediction models can be swiftly replaced.
COVID-19 has also expanded the potential uses of EVEscape. The pandemic provided important lessons about AI predictions — and an immense pool of data. With that wealth of information, EVEscape can make surprisingly accurate predictions about understudied viruses for which we have scarce data. That could prove immensely valuable in future outbreaks.
Tired of endless traffic jams, hefty bills, and the weight of climate guilt, city dwellers across Europe are ditching their cars for cargo bikes — and never looking back. Here’s why I became one of them.
Once the vehicle of choice for a select group of urban eco-geeks, cargo bikes are now rolling into the mainstream, helped in part by the rise of electric assist.
In many places, e-cargo bikes are simply a cleaner, greener, funner, cheaper, healthier, and, usually, faster way to transport yourself, your loved ones, and a bunch of stuff across the cityscape.
Growing up in South Africa, I had never seen, let alone ridden, a cargo bike before moving to the Netherlands earlier this year. So, curious to find out what all the fuss was about, I got in touch with one of the leading e-cargo bike brands, Urban Arrow, and they generously agreed to lend me one of their ‘Family’ models for a month. We loved it so much that at the end of the trial, we ended up buying our own and kicking our car to the curb.
Since then, I have cycled our bakfiets over 2000km across city streets, muddy forest paths, and sand tracks. Aside from the daily school run and trips to the supermarket, I’ve transported three bicycles stacked on top of each other, 120 kilograms of building materials, and a massive tree from the garden centre.
Now that I think of it, I have hauled more awkwardly-shaped items in the bakfiets (think lawnmower, wheelbarrow, cabinets, etc.) than I can recall. I’ve even given a panicked friend (plus two suitcases) a lift to the train station to prevent them from missing their flight. And all of this with (usually) a smile on my face.
More than enough space for the whole family.The Urban Arrow easily swallow a week’s worth of shopping — and the raincover keeps it all dry. Two bikes, one rider? No problem.Capable of hauling up to 125kg of stuff, the bakfiets was invaluable during our recent house renovation.
Enter the two-wheeled pickup truck
Cargo bikes are essentially any bike that can carry lots of stuff. They come in two main forms: the long-tail cargo bike, where goods and kids are transported on seats at the back of the bike, and the front-loader, where the cargo hold is slung out ahead of the rider. Front-loaders are often called ‘bakfiets’ (which directly translates to ‘box bike’ in Dutch) and are in many ways the quintessential cargo bike.
Bakfietsen were first popularised in the Netherlands and Denmark in the 1930s and 40s for delivering milk, bread, and other goods throughout the city— a trend which spread across the world.
But in the 1960s, they practically disappeared. The biggest culprit? The automobile. Faster, bigger, and hyper-fashionable, cars elbowed out practically all other forms of transport to become the top dogs of the city streets, leaving cargo bikes destined for the scrap yard.
But now, over half a century later, cargo bikes are making a comeback.
One of the biggest push factors has been the fact that our cities are getting cramped and polluted, with increasingly high costs of living. One of the biggest pull factors (getting people out of their cars and onto cargo bikes) has been the rise of electric assist.
With the help of an electric motor, cargo bikes become a whole lot easier and quicker to ride which makes them more practical. Add to that the sleek, lightweight, user-friendly design of modern-day cargo bikes and it’s easy to see why they are selling faster than traditional bikes and even other electric models.
In Germany, for instance, sales of e-cargo bikes skyrocketed from 37,000 in 2018 to 165,000 in 2022. The country is one of several that subsidise the cost of a cargo bike purchase.
While the Netherlands, Germany, and Denmark are undoubtedly leading the e-cargo bike charge, they are also taking off across Europe and further afield — from Ukraine to Uganda, and Nigeria to Canada.
Phil Marciniak from British Columbia, Canada, takes the expression “there is no bad weather, only bad clothing,” to another level. Here he stands beside his cargo bicycle fitted with a homemade plow for clearing bicycle lanes and sidewalks. Credit: TIMES COLONISTAcross Europe, mail and other goods are regularly delivered by e-cargo bike. Credit: PostNLTwo BBC journalists, Kate Vandy and Anna Holligan, have spent the last few years working on building a mobile broadcast studio and office on two wheels. Credit: BBCPlumber Martin Broer from Derby, UK, swapped his big white van for an e-cargo bike. Credit: BBCOur four-legged friends also love cargo bikes. Credit: Yuba
Testament to their surging popularity, there is even an International Cargo Bike Festival that has grown from a small group of cycling nerds in the small Dutch city of Nijmegen to an event attended by thousands of people in the country’s capital Amsterdam.
“Our goal is to inspire the world to stop using cars and vans and instead start using cargo bikes. That’s why we’ve declared the 2020s the Decade of the Cargo Bike,” say the festival’s organisers Jos Sluijsmans and Tom Parr.
One of the first brands to lead the electric cargo bike revolution was Amsterdam-based Urban Arrow. The company was founded in 2010 by Jorrit Kreek and Gerald van Weel, two cargo bike dads who were tired of constantly arriving at their destination with sweaty shirts sticking to their backs.
The first prototype of the Urban Arrow ‘Family’ ebike was envisioned by industrial designer Wytze van Mansum, and upon release it immediately won a Eurobike award, — and the rest, as they say, is history.
Urban Arrow sold over 10,000 units in 2019, which it claims makes it the world’s leading manufacturer of electric cargo bikes. Acquired by Dutch cycling conglomerate Pon four years back, the company has over 100 full-time employees and a worldwide network of more than 500 dealers.
The specs
The combination of a narrow, 2-wheeled base, strong Bosch motor, and a typically Dutch upright seating position makes the Urban Arrow surprisingly comfortable to ride, and nippy around town. The box is positioned very close to the ground giving the bike a low centre of gravity, making it surprisingly stable, even when carrying 125kg of cargo (max recommended capacity excluding the rider).
Once you get used to it it is just as easy as riding a normal bicycle, despite weighing 50kg and measuring 2.7 metres in length. The sheer size of the thing does take some getting used to, however, and often requires a not-so-elegant 3-point (or sometimes 10-point!) turn when parking.
The box is made from expanded polypropylene, the same material used on the inside of bicycle helmets. This envelopes your precious cargo with a soft cushion while being significantly lighter than other cargo bike boxes, which are typically made from wood or plastic. However, it is easy to damage, so if you’re looking for a bike exclusively for hauling freight, it is probably best to check out some of the more heavy-duty options.
The Family has a very smooth Enviolo continuously variable transmission gear hub. Instead of click-shifting from one gear to another while in motion, you can turn the shifter when you’re at a standstill. This makes kicking off a heck of a lot easier, especially as the motor kicks in almost instantly after take-off.
The drivetrain has four levels of assistance, which you can toggle on a small Bosch Intuvia display. The screen also shows the odometer, lights, and battery level. It’s all very intuitive and simple to use. Most of the time I just used ‘Turbo’, the highest setting.
The entry-level model, the Active Line-plus, is powered by a mid-drive Bosch motor supplied by a 400Wh battery that puts out 50Nm of torque. This is enough power for most applications — if you live in a flat country like the Netherlands that is.
If you’re tackling hills or hauling heavy cargo on the regular I’d highly recommend going for the Performance Line (65Nm) or the Cargo Line (85Nm) models. These also come standard with a larger, 500Wh battery that boosts the range to an acceptable 50km.
Charge time for all the batteries from nearly zero is about 4 1/2 hours. I ran out of battery just once over the past 2000-odd kilometres of ride time. That was a huge bummer though, and not something I (or my calves) would like to repeat.
The Urban Arrow Family starts at €5,290 for the Active Line, €5,790 for the Performance, and €6,690 for the Cargo. That’s not including essential accessories like a raincover. This is the price of a small, second-hand car in most countries, and is a significant investment whichever way you slice it.
Whether you think that the investment is worth it or not is a deeply personal decision. But when weighing up the pros and cons, consider this: ditching your car (or second) for a cargo bike is a lifestyle change that can have a profoundly positive impact on the health, wealth, and well-being of you, your family, your city, and your planet.
A gamechanger
The bakfiets is no ordinary bicycle. It has the potential to be a car replacement, or at the very least a substitute for your second car. And once you make the initial outlay, the running costs are minuscule in comparison. (The average European forks out around €300 per month to run, maintain, and insure their vehicle.)
Then there’s the fact that traversing the city by cargo bike is a lot of fun, especially in countries that have decent bicycle infrastructure. It forces you to travel at a slower pace, explore new parts of the city, and interact with the world around you in a more intimate way.
Even when doing the school run in the pouring rain, I find it hard to be grumpy when riding my cargo bike.
Taxi?
Cycling every day (which ditching your car for a cargo bike forces you to do) can work wonders for your physical health, too. During my one-month trial, I clocked up about 727km, which, according to Urban Arrow’s app, means I burnt an average of 200 calories per day. That’s more than the recommended daily amount of exercise to maintain a healthy heart, according to the John Hopkins Institute.
And then there’s the unequivocal fact that e-cargo bikes are better for the planet. Not only do they emit a fraction of the emissions during manufacture and operation as compared to other modes of transportation like cars, buses, and trains, cargo bikes also take up far less space. (Which makes it super easy to find parking, too).
In the overcrowded cities of today — where cars are increasingly impractical, expensive, noisy, and polluting — it’s no wonder that leaders across Europe and the globe are pushing to make their cities more bike-friendly.
Nevertheless, these benefits are owed to all bicycles. What really sets cargo bikes apart is their sheer practicality. Many of the items I have carried simply wouldn’t have fit in my old hatchback, and they also get the job done faster in many cases. My usual school run in the car used to take an average of 34 minutes there and back, based on a months-worth of data I collated using the fitness tracking app Strava. The cargo bike did it on average in 29 minutes.
This echoes a study from climate charity Possible which found that electric cargo bikes deliver 60% faster than vans in urban centres, totting up higher average speeds and dropping off ten parcels an hour, compared to a van’s six.
Well said Bewildered_Octopus on Reddit.
Making the shift
Despite all the myriad benefits of ditching your car for an e-cargo bike, it can be a big adjustment.
Naturally, cargo bikes will not cover the same distances that a car can — nor do they offer protection from the elements. They’re also a fair bit more expensive than a normal bicycle, and really big, which means you need somewhere secure to store them. For some, especially those who live in rural areas, they simply might not be a practical option. And even for those who are a perfect fit for the cargo bike life, some will simply prefer cars. In a way, I understand the sentiment — cars ooze comfort and convenience and the “need” for one is pretty much baked into our collective consciousness.
But most of the time, I simply don’t miss my car. I feel lighter, and more agile navigating my way about town. My kids absolutely love it, and it’s comforting knowing that I am producing zero carbon emissions and making the city a more liveable place for everyone.
Making the shift is more than just buying a cargo bike, however, it’s about viewing mobility in a different way. It’s about changing lanes, shifting gears, taking life at a different pace. Above all, it’s about valuing the journey over the destination. But trust me, once you do it, you won’t look back, and if you really miss driving a car — just rent one instead!
Signed, your cargo bike convert.
Becoming a cargo biker can be a big adjustment. Here are my top 5 tips to ease the transition:
Get proper rain gear — If you’re going to be using your bike a lot, in all seasons, get yourself a proper rain jacket, water-proof pants, and a solid pair of boots.
Accessorise — Most cargo bike brands offer a tonne of accessories, some more essential than others. The ones you can’t do without in my opinion are a good raincover, a floor mat, and a rack for the back (which allows you to mount another child seat or more bags for storage).
Sign up for a car-sharing app — Hate to burst your bubble, but, especially if you have kids, you’re probably going to need a car sometime or another. But that doesn’t mean you need to own one. When we need to go on longer trips we usually hire one, borrow from a friend or family member, or make use of the many car sharing scheme in our area.
Don’t forget to charge — If you do you’ll regret it, big time. Unlike a regular e-bike, cargo bikes are really heavy and you don’t want to be caught miles from home with a flat battery.
Beware crosswinds — Especially when you have the raincover on. Strong winds can hit the side of the cargo hold and push you into the road if you’re not careful. The best way to avoid this is to leave both sides of the cover a bit open so the wind can flow through.
Chatbots have an alarming propensity to generate false information, but present it as accurate. This phenomenon, known as AI hallucinations, has various adverse effects. At best, it restricts the benefits of artificial intelligence. At worst, it can cause real-world harm to people.
As generative AI enters the mainstream, the alarm bells are ringing louder. In response, a team of European researchers has been vigorously experimenting with remedies. Last week, the team unveiled a promising solution. They say it can reduce AI hallucinations to single-figure percentages.
The system is the brainchild of Iris.ai, an Oslo-based startup. Founded in 2015, the company has built an AI engine for understanding scientific text. The software scours vast quantities of research data, which it then analyses, categorises, and summarises.
Customers include the Finnish Food Authority. The government agency used the system to accelerate research on a potential avian flu crisis. According to Iris.ai, the platform saves 75% of a researcher’s time.
The <3 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
What doesn’t save their time is AI hallucinating.
“The key is returning responses that match what a human expert would say.
Today’s large language models (LLMs) are notorious for spitting out nonsensical and false information. Endless examples of these outputs have emerged in recent months.
Sometimes the inaccuracies cause reputational damage. At the launch demo of Microsoft Bing AI, for instance, the system produced an error-strewn analysis of Gap’s earnings report.
At other times, the erroneous outputs can be more harmful. ChatGPT can spout dangerous medical recommendations. Security analysts fear the chatbot’s hallucinations could even drive malicious code packages towards software developers.
“Unfortunately, LLMs are so good in phrasing that it is hard to distinguish hallucinations from factually valid generated text,” Iris.ai CTO Victor Botev tells TNW. “If this issue is not overcome, users of models will have to dedicate more resources to validating outputs rather than generating them.”
AI hallucinations are also hampering AI’s value in research. In an Iris.ai survey of 500 corporate R&D workers, only 22% of respondents said they trust systems like ChatGPT. Nonetheless, 84% of them still use ChatGPT as their primary AI tool to support research. Eek.
These problematic practices spurred Iris.ai’s work on AI hallucinations.
Iris.ai uses several methods to measure the accuracy of AI outputs. The most crucial technique is validating factual correctness.
“We map out the key knowledge concepts we expect to see in a correct answer,” Botev says. “Then we check if the AI’s answer contains those facts and whether they come from reliable sources.”
A secondary technique compares the AI-generated response to a verified “ground truth.” Using a proprietary metric dubbed WISDM, the software scores the AI output’s semantic similarity to the ground truth. This covers checks on the topics, structure, and key information.
Another method examines the coherence of the answer. To do this, Iris.ai ensures the output incorporates relevant subjects, data, and sources for the question at hand — rather than unrelated inputs.
The combination of techniques creates a benchmark for factual accuracy.
“The key for us is not just returning any response, but returning responses that closely match what a human expert would say,” Botev says.
Iris.ai founders (left to right) Maria Ritola, Jacobo Elosua, Anita Schjøll Abildgaard, and Victor Botev. Credit: Iris.ai
Under the covers, the Iris.ai system harnesses knowledge graphs, which show relationships between data.
The knowledge graphs assess and demonstrate the steps a language model takes to reach its outputs. Essentially, they generate a chain of thoughts that the model should follow.
The approach simplifies the verification process. By asking a model’s chat function to split requests into smaller parts and then displaying the right steps, problems can be identified and resolved.
The structure could even prompt a model to identify and correct its own mistakes. As a result, a coherent and factually correct answer could be automatically produced.
“We need to break down AI’s decision-making.
Iris.ai has now integrated the tech into a new Chat feature, which has been added to the company’s Researcher Workspace platform. In preliminary tests, the feature reduced AI hallucinations to single-figure percentages.
The problem, however, has not been entirely solved. While the approach appears effective for researchers on the Iris.ai platform, the method will be difficult to scale for popular LLMs. According to Botev, the challenges don’t stem from the tech, but from the users.
When someone does a Bing AI search, for instance, they may have little knowledge of the subject they’re investigating. Consequently, they can misinterpret the results they receive.
“People self-misdiagnose illnesses all the time by searching their symptoms online,” Botev says. “We need to be able to break down AI’s decision-making process in a clear, explainable way.
The main cause of AI hallucinations is training data issues. Microsoft recently unveiled a novel solution to the problem. The company’s new Phi-1.5 model is pre-trained on “textbook quality” data, which is both synthetically generated and filtered from web sources.
In theory, this technique will mitigate AI hallucinations. If the training data is well structured and promotes reasoning, there should be less scope for a model to hallucinate.
Another method involves removing bias from the data. To do this, Botev suggests training a model on coding language.
At present, many popular LLMs are trained on a diverse range of data, from novels and newspaper articles to legal documents and social media posts. Inevitably, these sources contain human biases.
In coding language, there is a far greater emphasis on reason. This leaves less room for interpretation, which can guide LLMs to factually accurate answers. On the other hand, it could give coders a potentially terrifying power.
“It’s a matter of trust.
Despite its limitations, the Iris.ai method is a step in the right direction. By using the knowledge graph structure, transparency and explainability can be added to AI.
“A wider understanding of the model’s processes, as well as additional outside expertise with black box models, means the root causes of hallucinations across fields can be sooner identified and addressed,” says Botev.
The CTO is also optimistic about external progress in the field. He points to the collaborations with LLM-makers to build larger datasets, infer knowledge graphs from texts, and prepare self-assessment metrics. In the future, this should yield further reductions in AI hallucinations.
For Botev, the work serves a crucial purpose.
“It is to a large extent a matter of trust,” he says. “How can users capitalise on the benefits of AI if they don’t trust the model they’re using to give accurate responses?”
The world’s largest offshore wind farm has started exporting power to the UK grid after its first turbine came online this weekend.
The Dogger Bank Wind Farm, currently under construction in the North Sea, will comprise a total of 277 turbines once complete in 2026. The huge plant is expected to churn out 3.6GW of power — enough for 6 million UK homes.
Prime Minister Rishi Sunak hailed the milestone as a boost to energy security and job creation. “Offshore wind is critical to generating renewable, efficient energy that can power British homes from British seas,” he said.
Dogger Bank is an attractive location for offshore wind farms because it is located far away from shore, avoiding complaints about the visual impact of wind turbines. The scheme is central to the government’s plan to ramp up offshore wind capacity from 13.7GW to 50GW by 2030 — enough energy to power every home in the country.
China (49%), the United Kingdom (22%), and Germany (13%) currently account for more than 75% of global offshore wind installed capacity.
While positive news for the UK’s energy transition, the powering up of Dogger Bank comes amid a watering down of climate policy by the country’s Conservative government.
Restrictions on onshore wind and solar
In the latest move against renewable energy projects, Sunak is planning to clamp down on solar panel installations across British farmland, which campaigners say will stall the UK’s green transition and raise bills, the Observer reported this week.
Under the plans, solar farms could be blocked by local authorities if officials believe a prospective project could “put food security at risk.” Downing Street is expected to cite food shortages in Europe and the war in Ukraine to justify the decision, with environment secretary Therese Coffey also reportedly backing the move.
The proposal will no doubt be welcomed by many Conservative MPs and rural residents, who have been campaigning against new solar farm developments, such as the proposed Sunnica Energy Farm, for years.
Despite concerns, ground-mounted solar panels currently cover just 0.1% of all land in the UK. Even government plansto significantly scale up solar in line with its net-zero target are expected to bring this up to just 0.3% of the UK land area.
“Solar energy will help us move away from polluting fossil fuels, and in the long term protect UK farming from climate breakdown. Restricting ground-mounted solar would be gravely short-sighted,” Lydia Collas, senior policy analyst at Green Alliance, told the Guardian.
Some even took to social media to express their dismay at the proposal. “I guess Rishi’s rich landowner mates don’t want their country views spoiled,” said one commenter on Reddit. Another user rightly pointed out that you can easily farm and generate power on the same land — a system known as agrivoltaics.
The plans threaten to stifle the growth of solar power in the country by making it harder for new projects to gain planning permission.
It follows an almost decade-long policy that effectively banned the installation of new onshore wind turbines, one of the cheapest sources of electricity currently available.
The UK government’s aversion to onshore wind dates back to David Cameron’s stance in 2015, which Sunak vowed to uphold when he ran for the Tory leadership last year. Sunak said at the time that he instead wanted to pursue offshore wind due to the “distress and disruption” onshore wind farms can cause to local residents.
Even though the de facto ban was relaxed last month, onshore wind in England still faces higher planning barriers than anything else, including new coal mines, according to climate advocacy group Possible.
The point of all of this is that British policymakers are effectively stalling the energy transition by tying up key onshore renewable energy projects in the slow (and often right-leaning) legislative processes of local authorities.
This seemingly contradicts the UK’s future energy commitments, which aim to ramp up onshore wind and solar from, respectively, 14.5GW and 13.8GW today to 35GW and 53GW by 2035.
Sunak’s anti-green agenda
Unfortunately though, the restrictions on new onshore wind and solar imposed by the UK government are pretty on brand.
In a speech on September 20, Rishi Sunak promised a “new approach” that in effect significantly waters down Britain’s climate policy.
In that speech, Sunak announced that the 2030 phase-out of new petrol and diesel cars will be pushed back to 2035, as well as weakening the 2035 gas boiler phase-out, confirming it will apply to far fewer homes.
Moving too fast on green policies, he said, “risks losing the consent of the British people.”
The move came just two months after the UK controversially granted 100 new licenses for oil and gas exploration and production in the North Sea.
Sunak has since faced mounting criticism for his alleged anti-green agenda, including from those within his own party.
Former US Vice President Al Gore described the changesas “shocking and disappointing” and “not what the world needs from the United Kingdom.” Britain’s weakening of climate policies comes at a time when extreme weather events are becoming more frequent across the world, including the UK, as a result of global heating. Only through the swift rollout of a variety of renewables can we avert even greater disaster, urge scientists.
Today, France’s Eviden (part of cybersecurity, cloud, and high-performance computing group Atos) and German modular supercomputing company ParTec, announced they had won a contract to provide the very first exascale supercomputer in Europe.
The JUPITER project will cost €500mn in total. The computer itself will cost €273mn and run on Arm architecture SiPearl Rhea processors and Nvidia accelerator technology. It will be operated by the Jülich Supercomputing Centre in Germany.
JUPITER will be the first system in Europe to surpass the threshold of one billion billion calculations per second. The aim is for it to support the development of high-precision models of complex systems, which could help solve questions regarding areas such as climate change, pandemics, and fusion energy. Of course, it would also enable intensive use of AI and analysis of large data volumes.
JUPITER stands for Joint Undertaking Pioneer for Innovative and Transformative Exascale Research (just in case you were wondering). The European High Performance Computing Joint Undertaking (EuroHPC JU) announced the project last year, and put out a call for tenders in January.
But let’s back up for a moment — what exactly is an exascale supercomputer?
One billion billion flops per second
An exascale system, as already mentioned, is a supercomputer or high-performance computing (HPC) system capable of performing a billion billion calculations per second. This is equivalent to one exaflop.
In other words, an exaflop is a measure of performance for a supercomputer that can calculate at least one quintillion (exa-) floating point operations (flop) per second. Meanwhile, an exabyte is a memory subsystem packing a quintillion bytes of data.
Building and operating exascale systems pose various technical challenges, including power consumption, heat management, scalability, and software optimisation.
The world’s first exascale supercomputer is the Frontier, built by Hewlett Packard Enterprise (HPE) and housed at the Oak Ridge National Laboratory in Tennessee. It was deployed in 2021 and reached full capacity in 2022. It is set to be superseded by El Capitan at the Lawrence Livermore National Laboratory in California, also by HPE. El Capitan will deliver over 2 exaflops when it comes online mid-2024.
Meanwhile, the fastest supercomputer in Europe, owned by the EuroHPC JU, is the Lumi (Large Unified Modern Infrastructure). It sits in the CSC data centre in Kajaani, Finland, began operating in 2021, and can achieve more than 375 petaflops (one thousand million million flops per second), with a “theoretical peak” at 550 petaflops. That makes it the third fastest supercomputer in the world as of June 2022.
In the latest round of geopolitical tech chess, the European Commission today published a list of critical technologies to keep safe from geopolitical rivals, in an effort to bolster the bloc’s own economic (and not only) security.
The document, prepared by the Commission’s digital, defence, and trade chiefs in consultation with the member states, will serve as the basis for an outgoing investment and export control tool.
The list consists of 10 technologies, with applications ranging from potential human rights violations to military robots, quantum supremacy, genetic modification, and interstellar travel. The Commission reportedly considers four to be particularly dangerous, should they fall into the wrong hands. These are:
Biotechnologies (techniques of genetic modification, new genomic techniques, gene-drive, synthetic biology).
The remaining six technologies are advanced connectivity and navigation; advanced sensing technologies; space and propulsion technologies; energy technologies (including fusion and hydrogen); robotics and autonomous systems; advanced materials manufacturing and recycling.
The bloc’s industry chief, Thierry Breton, stated that Europe was “adapting to the new geopolitical realities, putting an end to the era of naivety and acting as a real geopolitical #power.” Make of that use of a hashtag what you will.
Today we are de-risking our economy by identifying 10 areas of #CriticalTechnologies⤵️
Europe is adapting to the new geopolitical realities, putting an end to the era of naivety & acting as a real geopolitical #power. pic.twitter.com/Q9Rt1nkfoU
While China is not explicitly referred to in the document, sources familiar with the matter told Politico that it was “like the EU’s Voldemort. The country that cannot be named.”
The new era of soft power tech
Ever since the beginning of civilisation, technology has defined geopolitics. Think marauders rumbling over Ancient Egypt in chariots made possible by the invention of the wheel, or, much closer in time, the principle of mutual assured destruction (MAD). With globalised supply chains and the rise of digitalisation, it is not as simple as one actor has the tech while the other doesn’t, and thus gains the upper hand. Never before has the intersection of geopolitics and technology been so intricate.
The battle for semiconductors that power so much of not only our day-to-day existence but also military technology around the world will only intensify as AI requires more and more powerful chips. In addition, the last few years have seen a growing number of cyberattacks on government agencies and service providers, while some authorities turn to digital technology to surveil their own people.
Digital sovereignty and “de-risking” of economic policies have risen high on the political agenda, along with mitigating rising geopolitical technological threats — without disturbing sensitive supply chains. One way of doing this is through export controls. In August, US President Joe Biden unveiled an executive order banning new investments in Chinese tech sectors related to artificial intelligence and quantum computing. This was merely the latest move in a tech trade back-and-forth spanning several months, with China restricting exports of two key semiconductor minerals in July on grounds of ”‘national security.”
Both the EU and the US have been busy trying to shore up domestic semiconductor and chipmaking capabilities. Germany in particular has been eager to throw substantial amounts of cash at the problem. However, it still might not solve one of the most immediate key issues for either — a lack of highly skilled engineering talent. Still, they are looking ahead and playing the semiconductor long game.
Relationship with Beijing characterised by low levels of trust
While neither the EU nor the US is likely to establish independence, doing as best one can now might well turn out to be prudent. The global chip industry is very much dependent on one company — Taiwan’s TSMC. It produces close to 60% of the world’s computer chips, and about 90% of the most advanced ones.
Taiwan is under increasing geopolitical pressure from China. For instance, its imposing mainland neighbour has been performing military exercises simulating a blockade of the island state, and President Xi has openly declared it is his generation’s obligation to seek reunification.
While tensions in the South China Sea are undoubtedly rising, it need not even come to military confrontation for the West to lose access to TSMC’s mega fabs. Look at the political developments in Hong Kong over the past few years — it might only take the “election” of a Beijing-friendly government in Taiwan.
In the words of Commission Vice President Věra Jourová when commenting on the EU’s forthcoming landmark AI Act, “We see this as setting the standards for the democratic world but China is not part of this because in the world of technologies we are more rivals than partners.”
Jourová further added that there was a “low level of trust” between Beijing and the remainder of the G7 countries set to meet in Kyoto next week to discuss the evolution of AI technology. Indeed, it seems the age of tech trade diplomacy has only just begun. Hopefully, the inevitable conflicts on the road ahead can be solved by just that — diplomacy.
The European Commission has been too focused on tackling Big Tech and not enough on improving digital services for citizens.
That is according to Poland’s minister of digital affairs, Janusz Cieszyński. He believes the EU needs to shift its digital policy focus to helping Europeans interact with government services.
Speaking to TNW at the recent Tallinn Digital Summit in Estonia, Janusz Cieszyński said that digital ID services that are prevalent in Estonia and increasingly so in Poland under his remit, need to be a higher priority EU-wide.
“There’s a lot of focus on [tech businesses] and there’s not enough focus on legislation which will help citizens, which will help ordinary people for instance to travel around Europe with your ID in your phone,” he said.
“When there was the difficult time of the pandemic, we were able to roll out the Covid certificates very quickly and now we have fallen back asleep, and I don’t think that things are going fast enough.”
Cieszyński leads Poland’s digitisation ministry, including its mObywatel (mCitizen) initiative, which provides a mobile app for users to store their ID documents. The ministry has also updated the app to allow Ukrainian refugees in Poland to upload and store their documents.
He said that while there’s a slew of digitisation efforts across member states, there should be more cohesion on an EU level.
Thierry Breton, the European Commissioner for the Internal Market, who delivered a keynote speech at the summit in Tallinn, championed many of these efforts.
This includes the Digital Services Act and the Digital Markets Act, two of the European Commission’s flagship tech regulations that it believes will tackle misinformation on social media and create fairer competition in Europe.
“I’m not sure that as a consumer I have seen an improvement during Mr Breton’s tenure as Commissioner,” Cieszyński said.
“It is obvious to me as a person who has supervised numerous rollouts of e-services, that if you would be committed to having an e-ID which would be universal for all European countries and spend five years on this, we would already have been using this for quite some time,” he said.
“I think it’s just [that] the Brussels bureaucracy is too detached from the citizens and their base. I think that’s the thing.”
A spokesperson for the European Commission declined to comment on the minister’s remarks.
Upcoming EU elections may hinder progress
The EU has debated and proposed a unified approach for digital IDs several times over the years.
The Commission proposed legislation for a secure digital ID system in 2021 that would be in effect by 2030.
In June this year, EU member states and the European Parliament reached an agreement on an EU-wide framework for a digital ID that would be interoperable around the bloc.
Now the proposal must weave its way through the usual EU law-making hurdles. With the European Parliament elections next year coinciding with the end of the current Commission’s mandate, it is difficult to affix a timeline for the intended legislation.
The Commission spokesperson did tell TNW that the framework would provide Europeans with “consumer control, security, convenience and privacy.”
“These common European rules will ensure that solutions can be used seamlessly cross-border, also creating new business cases for companies,” they said. “The Commission already started the technical groundwork with Member States to ensure this interoperable, secure, and user-friendly digital personal wallet becomes a reality for citizens.”
The European Commission does have a series of projects involving member states, universities, and private sector companies underway to examine various digital ID use cases in travel, payments, healthcare, and education.
Examples include the European Wallet Consortium (EWC), which is testing interoperable IDs for travel, and Digital Credentials for Europe (DC4EU), which is working with several European universities and government departments to examine how a digital ID for the education sector could work.
E-government services
Estonia is often held up as a bastion of how to implement digital services.
Since the ‘90s, the Baltic state has developed a digital-first approach to government services with most functions now available online. Most notably, citizens can vote online in elections; in fact, in the 2023 parliamentary elections more than half of the votes were cast over the internet. A forthcoming change in Estonia’s digital policies will allow for the completion of divorce proceedings online.
But pushing that agenda can be challenging.
The government has contracted Estonian IT services firm Nortal to build several of its digital public services.
Chief growth officer Elizabeth Kiehner said that the company is trying to “export” its technology to other countries but there are numerous obstacles to overcome.
“The prerequisite to any of these things working is having a leader with the political will. There could be very little that’s currently digitised but there needs to be a leader with political will and commitment to do this,” she said.
Competing for tech talent
For governments with ambitions to expand digital services, one of the biggest challenges is finding the tech talent it needs.
Cieszyński told TNW that governments must compete for tech talent against the private sector without being able to offer the higher salaries typically expected at bigger tech companies.
It is common for public bodies to outsource development to private companies. However, Cieszyński believes that it is imperative for governments to bring as much tech development in-house as possible.
Government departments and bodies need to know what’s in the “black box” and understand what’s under the hood and how it works.
“You have to remember that this is software that will be used for the processing of government data, of citizen data, super sensitive stuff, and it is also, something which comes to mind very easily in these difficult times, a matter of security. If you don’t know what’s inside this black box, how can you be sure that your citizens, the ones that you’re supposed to protect, are actually protected?” Cieszyński explained.
“We work a lot with vendors. We use modern technology, we buy from the top brands from around the world but we want to have a good proportion of experts that are in-house on our payroll, so that we have the rights to whatever code they write, things like that.”
Currently, Cieszyński’s department has 1,400 people working on e-government services.
Will political mandate cycles shift priorities?
The minister stated that e-government and digital IDs should be front of mind for the next European Commission.
Cieszyński is from Poland’s ruling PiS party, which faces a general election this month where the complexion of the Polish government could change. With that, so could priorities when it comes to digital policies.
As for the European Commission, its mandate is up in less than a year. Rumours abound that Thierry Breton will be seeking the top job of Commission President.
But just what will the next Commission’s priorities be?
During his keynote speech in Tallinn, Breton dropped one or two hints when talking about supercomputers and quantum.
“One of my favourite subjects, quantum. Of course, we are working very hard on quantum computing, quantum communication, quantum sensing and cryptography,” he said. “This is something we will push hard and will be a very interesting subject for the next Commission.”
The UK, Switzerland, and Sweden are poised to cash in from the AI gold rush, but most of Europe will be a poor nephew to Uncle Sam.
So say the expert analysts at Capital Economics, a financial research firm based in London. In a new study, the company assessed which countries are best placed to benefit from the AI boom — and which ones will be left behind.
Using 40-sub-indicators, the researchers analysed their capacities for AI innovation, diffusion of the AI effectively, and adaptability to the impacts. The three categories were then combined to calculate a global ranking.
Unsurprisingly, the US topped the charts, but there were some shockers in the chasing pack.
Among them was the relatively low position of China. Despite the phenomenal AI developed at companies like TikTok and within research institutions, immense regulatory barriers and government intervention in the private sector sunk China to a middling performance.
The <3 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
In Europe, meanwhile, the outlook is mixed. Leading the continent is the UK, which ranked third globally, behind only the US and Singapore.
“The UK could benefit from lighter-touch regulation.
The study team gave several reasons for Britain’s loft position. Despite perennially low investment rates, the country has become a magnet for AI talent and boasts a propitious business foundation.
One of the country’s greatest strengths is its higher education system, particularly in the golden triangle formed by the university cities of Cambridge, London, and Oxford, which attract top talent from around the world.
Another innovation engine is Google DeepMind’s pioneering AI lab, which is based in London. The analysts also expect the UK’s flexible labour market to support the wider economic adaption to AI, while its service-oriented economy can accelerate the diffusion.
Britain may be further impacted by its position outside of the EU — both positively and negatively. While Capital Economics doesn’t expect Brexit to be a major factor in the AI economy, the research team can envision some effects.
“On the plus side, the UK could benefit from lighter-touch regulation if the EU regulates AI more strictly in future,” Andrew Kenningham, the firm’s Chief Europe Economist, told TNW. “But set against that, there could be reduced collaboration on AI projects or the UK could be hindered from participation in large-scale European AI initiatives.”
Next in line for the throne
The UK is one of three European countries rounding out the global top five. Taking fourth spot is Switzerland, while Sweden bagged fifth. Both countries were particularly strong in adaption, ranking first and second in the world, respectively.
“This essentially means that they have a good track record on redeploying resources,” Kenningham said.
In terms of innovation, Switzerland and Sweden ranked slightly higher than Germany and a bit lower than the UK. But they both lagged far behind China and the US — a common issue in Europe.
“Europe is a long way behind the US and China on our sub-index of ‘innovation’ essentially because it invests less in AI research and has less academic research in that area,” Kenningham explained.
Composite scores of both the Euro-zone and the emerging economies in the EU are behind those in the US and Asia. Credit: Capital Economics
Outside of the continent’s three leaders, the European landscape is considerably bleaker.
One of the gloomiest areas is a scarcity of financial backing. Investors are ploughing enormous capital into US tech firms with GenAI capabilities, whereas European equities are receiving a lesser AI-related boost.
As a result, the continent’s stock markets will likely struggle to keep pace with their counterparts across the Atlantic. These divides will be amplified if the US market entices more European firms, as it recently did to Arm, the UK-based chip designer.
A further obstacle is the continent’s sparse cloud infrastructure, which provides essential foundations for AI. There also remain several enduring barriers. Compared to the US, Europe has a small VC ecosystem, scarce enormous companies, rigid labour markets, tight regulations, and strict local planning restrictions. Collectively, these components restricted growth during the 1990s ICT boom. In the AI era, they may create even greater hindrances.
To emphasise this point, the study team note the impacts of EU tech regulations. A prominent example recently emerged with the release of Bard, Google’s answer to OpenAI’s GPT. Due to concerns over GDPR compliance, Bard arrived in over 100 countries before the EU.
Future launches will be further inhibited by the impending AI Act, which restricts technologies such as biometric surveillance and emotion recognition. Protecting the public doesn’t always please businesses — or the economy.
The future is not set
There’s still time to move up — or down — the rankings. As with previous transformative technologies, AI’s productivity boost will likely be more of a slow-burn than an overnight surge. Capital Economics expects the big impacts to come in the late 2020s and 2030s.
To improve their preparations, the firm advises European governments to more actively support immigration of IT and AI talent. The researcher also championed the UK’s provision of fiscal aid for academic and commercial research.
In the long-term, further support for the enabling environment, such as universities and education in STEM subjects, will also have a positive impact.
“Apart from innovation/R&D governments can also do more to support the diffusion of AI through, for example, ensuring that regulations do not discourage adoption, labour markets are flexible, and [the] public sector leads by example in adopting new technology,” Kenningham said.
Without these changes, much of Europe is set to slip further behind the US in the global economic order.
European Union bureaucracy might not conjure the most exciting of connotations. However, being part of the “infinity team” surely puts a superhero-esque spin on your average Frankfurt grey high-rise working day.
Minute takers watch out. After surveying employees on where deploying generative AI could be most effective, the ECB’s newly established working group has launched nine trials, the results of which could speed up day to day activities of the financial institution.
Large language models, the organisation says, could be deployed for tasks including writing draft codes, test out software faster, summarising supervision documents, drafting briefings, and “improving text written by staff members making the ECB’s communication easier to understand for the public.”
The Central Bank’s chief service officer, Myriam Moufakkir, discussed the ECB’s use of AI on Thursday, in an organisation blog post. (To be perfectly straight, she doesn’t mention the “infinity team” by name, but other reports submit this to be the assigned designation.)
Commenting on the ECB’s core work of analysing vast amounts of data, Moufakkir said, “AI offers new ways for us to collect, clean, analyse, and interpret this wealth of available data, so that the insights can feed into the work of areas like statistics, risk management, banking supervision, and monetary policy analysis.”
Existing ECB applications of AI
Moufakkir said that the ECB is already deploying AI in a number of areas. This includes data collection and classification, as well as applying web scraping and machine learning to understand price setting dynamics and inflation behaviour.
Furthermore, it applies natural language processing models trained with supervisory feedback to analyse documents related to banking supervision. This is done through its in-house Athena platform, which helps with topic classification, sentiment analysis, dynamic topic modelling, and entity recognition. It also uses AI to translate documents into member state languages.
“Naturally, we are cautious about the use of AI and conscious of the risks it entails,” Moufakkar further noted. “We have to ask ourselves questions like ‘how can we harness the potential that large language models offer in a safe and responsible manner?’, and ‘how can we ensure proper data management?’.”
Without specifying exactly how, she added that the ECB was looking into “key questions” in the fields of data privacy, legal constraints, and ethical considerations. With the EU’s landmark AI Act in the works, the use of financial data of its citizens should be a particularly intriguing landscape to navigate.
AI has thrust deeper into carnal pleasures with the launch of a new “roleplaying” chatbot system.
The feature is the brainchild of Bloom, an erotic audio platform based in Germany. The app also comes with 800 original sex stories, which range from the vanilla to the kinky (anyone for sploshing?).
Some of the tales offer guided erotic play, while others feature creator content from real couples. There are even stories that synchronise with the vibrations of sex toys.
The new chatbots add another level of interaction. Hannah Albertshauser, Bloom’s CEO, says the feature takes the character-driven storytelling “to the next level.”
“Our unique approach captivates the mind more than just the eyes,” Albertshauser tells TNW.
The <3 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
As a selfless service to our faithful readers, TNW put her claim to the test.
First dates
Ahead of the launch, TNW received a demo with a choice of seven characters: Lizzy, Noah, Barry, Noah as the Architect, Joel, Fleur as Mistress Mia, and Joel as the Ranch Hand.
According to Bloom, the underlying AI models are trained to represent the charm, personality, and language of each character. Rob the Ranch Hand, for instance, may call you “darlin’,” while Mistress Mia could ask what pet name you like to be called.
Eagle-eyed readers may have noticed that most of the personas are male. Albertshauser said this reflects an “overwhelming desire” from users to connect with those specific individuals.
That desire may also mirror broader consumer tastes. Studies have consistentlyshown that sex stories are particularly popular with women, whereas men are more likely to watch porn (more on that later).
Albertshauser promises that additional “strong female characters” will soon make appearances. But TNW felt sufficiently indulged with seven virtual sweethearts at our beck and call.
In an interface that resembles dating apps, we slide into the DMs of the AI-powered characters. An amorous opening message then asks about our fantasies to set the tone of the conversation.
After entering a prompt, the chatbot replies with personalised voice and text messages. As the questions and responses flow, the feature engenders an immersive and customisable experience.
Noah is one of Bloom’s most popular characters. Credit: Bloom
We’ll leave the juicy details to your sordid imaginations, but we were impressed by the personalised and natural interactions. The most striking aspect was the integration of details from user prompts.
“The more dynamic and customisable an experience can be, the more a user’s exact preferences and desires can be satisfied,” Albertshauser says.
“This is what we set out to change by introducing a feature where users can tell their favourite voices exactly what they are in the mood for and what fantasies they want to explore.”
Those voices may, however, be seen more than they’re heard. While a free version provides unlimited text responses, the audio messages are only available to premium users. Subscribers receive 15 of them per month. Additional voice message tokens can then be purchased in the app.
“We believe Bloom chat can strengthen real-world connections.
To determine the format of each response, Bloom developed an algorithm that selects either a text or audio response. Albertshauser says that it takes “several factors” into consideration.
“There is also an element of chance to keep the voice message output randomised in a way that is exciting for the user and makes the voice messages feel special,” she adds.
At times, the system did decline our seedier requests. According to Bloom, the chatbots will not engage in anything that breaches the platform’s community guidelines.
The characters can also integrate consent and safewords to ensure the user’s comfort. Another feature, which was still being built at the time of our demo, will let users report individual messages.
These safety features are just the tip of Bloom’s plan to promote positive sexual interactions.
Safe sex
Bloom’s chatbots are designed for inclusive and consensual adult interactions. They aim to set realistic standards for pleasure, which cater to a wide range of audiences and sexualities.
The erotica offers a different experience to visual pornography, which can skew perceptions of sex, relationships, and boundaries. Mainstream porn also predominately caters to the male gaze. Female pleasure is often relegated to something that only exists to please men.
Bloom introduces a welcome alternative. In a similar manner to podcasts, audio erotica creates an intimacy between the speaker and listener. The chatbots make the connection even closer.
“The interactive side of Bloom chat, really feeling like you are the main focus, is what separates us from video pornography and something that is very valued by our users,” Albertshauser says.
The feature could also serve as a confidence-building tool. Albertshauser hopes people use the chatbots to become more comfortable exploring and expressing their sexual interests.
“We believe Bloom Chat can actually help strengthen real-world connections. As our users chat openly with the bots about their sexual desires and fantasies, it strips away the shame and fear many people feel when talking about these things with partners, and helps users practise these conversations before bringing them into the real world.”
Bloom also promises that all user data is anonymised and never shared with third parties or used for other purposes.
Bloom was cofounded by Hannah Albertshauser and her husband Mike. Credit: Bloom
To substantiate its claims about the feature’s benefits, Bloom sought input from Suzannah Weiss, a certified sex educator.
Weiss responded with a ringing endorsement.
“The characters are very respectful and affirming of your desires, which could make this function helpful for people who are overcoming sexual trauma, relationship abuse, or body image issues,” she says.
“It provides an example of what a positive sexual interaction looks like! I left the conversation with my chosen character feeling uplifted and excited to try out some of the fantasies we discussed with a live person.”
Happy endings
Bloom has also vowed to mitigate a common concern about AI: the danger of machines replacing human workers.
The training data for the chatbots includes the scripts and voice profiles of every character. It also contains AI voice-training materials recorded by each actor.
To protect their livelihoods, the performers could “opt-in” for a licensing model that provides compensation for their work. Once the terms were agreed, they were consulted to ensure the chatbots accurately represented their characters.
Bloom assures the voice actors, writers, and illustrators that the AI will not diminish their roles. Instead, the company says the feature will amplify their work, as the chatbots are modelled on characters and narratives that they created. Stories crafted and curated by humans will still be the platform’s core.
Those pledges won’t allay every concern about tech’s impact on relationships. The rise of deepfake porn, sex robots, and virtual girlfriends has aroused alarm that future sex will become inhuman. But Bloom’s chatbots show AI can at least offer artificial intimacy.
