Meta just announced a major move in its AI efforts--investing in Scale AI and building a superintelligence AI research lab. While Meta has been trying to keep up with big names in the AI race, such as OpenAI, Anthropic and Google, the company's new strategy includes dropping some serious cash to acquire talent and invest in Scale AI. Today on the show, we dive into the deal between Meta and Scale AI, including what Meta aims to get out of investment, and we ask the question we are all wondering: What is superhuman intelligence, anyway? Write to us at uncannyvalley@wired.com. You can always listen to this week's podcast through the audio player on this page, but if you want to subscribe for free to get every episode, here's how: If you're on an iPhone or iPad, open the app called Podcasts, or just tap this link.

Fox News correspondent Stephanie Bennett has the latest on the aftermath of the Kazakhstan plane crash on'Special Report.' Crew members and survivors of the Azerbaijan Airlines plane that crashed in Kazakhstan on Christmas Day say they heard at least one loud bang before the aircraft crashed in a ball of fire, heightening speculation that a Russian anti-aircraft missile may have been responsible for the tragedy. The Embraer 190 passenger jet flying from Azerbaijan to Russia crashed near the city of Aktau in Kazakhstan after diverting from an area of southern Russia where Moscow has repeatedly used air defense systems against Ukrainian attack drones. At least 38 people were killed while 29 survived. Subhonkul Rakhimov, one of the passengers aboard Flight J2-8243, told Reuters from the hospital that he had begun to recite prayers and prepare for the end after hearing a bang.

Generalization to out-of-distribution (OOD) data is a critical challenge in machine learning. Ensemble-based methods, like weight space ensembles that interpolate model parameters, have been shown to achieve superior OOD performance. However, the underlying mechanism for their effectiveness remains unclear. In this study, we closely examine WiSE-FT, a popular weight space ensemble method that interpolates between a pre-trained and a fine-tuned model. We observe an unexpected phenomenon, in which WiSE-FT successfully corrects many cases where each individual model makes incorrect predictions, which contributes significantly to its OOD effectiveness. To gain further insights, we conduct theoretical analysis in a multi-class setting with a large number of spurious features. Our analysis predicts the above phenomenon and it further shows that ensemble-based models reduce prediction errors in the OOD settings by utilizing a more diverse set of spurious features. Contrary to the conventional wisdom that focuses on learning invariant features for better OOD performance, our findings suggest that incorporating a large number of diverse spurious features weakens their individual contributions, leading to improved overall OOD generalization performance. Empirically we demonstrate the effectiveness of utilizing diverse spurious features on a MultiColorMNIST dataset, and our experimental results are consistent with the theoretical analysis. Building upon the new theoretical insights into the efficacy of ensemble methods, we further identify an issue of WiSE-FT caused by the overconfidence of fine-tuned models in OOD situations. This overconfidence magnifies the fine-tuned model's incorrect prediction, leading to deteriorated OOD ensemble performance. To remedy this problem, we propose a novel method called BAlaNced averaGing (BANG), which significantly enhances the OOD performance of WiSE-FT.

We propose a neuromorphic architecture for real-time processing of acoustic transients in analog VLSI. We show how judicious normalization of a time-frequency signal allows an elegant and robust implementation of a correlation algorithm. The algorithm uses binary multiplexing instead of analog-analog multiplication. This removes the need for analog storage and analog-multiplication. Simulations show that the resulting algorithm has the same out-of-sample classification performance (-93% correct) as a baseline template-matching algorithm.

ChatGPT is an amazing Chatbot with the ability, competency, and confidence to construct paragraphs, sentences, essays, and more. However, this optimism can be deceiving because it succumbs to several common misconceptions among the general public, even though experts know much better. Can an astrophysicist teach ChatGPT to acquire new information and absorb it such that it gives exact responses when it previously gave confident but mistaken ones? When we try to understand anything more intensely, we all end up in an awkward situation: we believe we know how something performs, only to explore that we are wrong. This entails not just learning what is true, but also why what we assumed was true was, in actuality, false, and how we can avoid making the same mistakes again.

We propose a routing algorithm that takes a sequence of vectors and computes a new sequence with specified length and vector size. Each output vector maximizes "bang per bit," the difference between a net benefit to use and net cost to ignore data, by better predicting the input vectors. We describe output vectors as geometric objects, as latent variables that assign credit, as query states in a model of associative memory, and as agents in a model of a Society of Mind. We implement the algorithm with optimizations that reduce parameter count, computation, and memory use by orders of magnitude, enabling us to route sequences of greater length than previously possible. We evaluate our implementation on natural language and visual classification tasks, obtaining competitive or state-of-the-art accuracy and end-to-end credit assignments that are interpretable.

"Ye shall know them by the temporal dynamics of their temperature parameter." God created the universe by strategically forgetting about something that would take a whole reality to remember. It doesn't mean He (gender neutral) did it on purpose. Everything that can happen, happens. See, the Hindus thought that the "way out" was by "remembering who you are".

Earlier this year, at the famed La Romareda stadium in Zaragoza, Spain, EA Sports organised two football matches, one each for male and female pro players. During these competitive 90-minute fixtures, all participants, including subs and officials, wore advanced Xsens motion capture suits that recorded their every movement, shot, tackle and celebration. Involving more than 70 people it was, according to gameplay producer Sam Rivera, the largest number of players ever motion-captured in a single session. Every year, the developers of Fifa tell us that their key aim is authenticity. This year, Fifa 23 – the final product of EA Sports and Fifa's 30-year partnership – is about making key moments more intelligible, detailed and dramatic, zooming in ever closer to the action at pitch level.

People are advocating for a kinder, more considerate approach to relationships. Simply put: humans are showing us their best side. The "others" they're defending and romancing are chatbots. I recently stumbled across the "Replika AI" subreddit where users of the popular chatbot app mostly congregate to defend the machines and post bot-human relationship wins. Tickets to TNW 2022 are available now! Users appear to run the gamut from people who genuinely seem to think they're interacting with an entity capable of actual personal agency: Of course, most users are likely just curious and enjoying the app for entertainment purposes.

As new year approaches, crowds around the world may be expecting whizzes and bangs to light up the sky. One notable example was the opening ceremony of this year's Tokyo Olympics, while the Over the Top NYE event at Reunion Tower in Dallas is among those planning to combine fireworks and drones to welcome 2022. They are also being embraced at a local level: more than 1,000 people watched a drone display at Mercia Marina in Derbyshire to celebrate Bonfire Night this year. Ollie Howitt, the creative coordinator at SkyMagic, which used a fleet of 300 drones to create a display for the mayor of London's new year celebration last year, said demand had increased substantially, something the pandemic has helped accelerate. She added that drones were increasingly able to fly in greater density and for longer.