Keeping up with an industry as fast-moving as\u00a0AI<\/a>\u00a0is a tall order. So until an AI can do it for you, here\u2019s a handy roundup of recent stories in the world of machine learning, along with notable research and experiments we didn\u2019t cover on their own.<\/p>\n This week in AI, Microsoft unveiled a new standard PC keyboard layout<\/a> with a \u201cCopilot\u201d key. You heard correctly \u2014 going forward, Windows machines will have a dedicated key for launching Microsoft\u2019s AI-powered assistant Copilot<\/a>, replacing the right Control key.<\/p>\n The move is meant, one imagines, to signal the seriousness of Microsoft\u2019s investment in the race for consumer (and enterprise for that matter) AI dominance. It\u2019s the first time Microsoft\u2019s changed the Windows keyboard layout in ~30 years; laptops and keyboards with the Copilot key are scheduled to ship as soon as late February.<\/p>\n But is it all bluster? Do Windows users really\u00a0want<\/em> an AI shortcut \u2014 or Microsoft\u2019s flavor of AI period?<\/p>\n Microsoft\u2019s certainly made a show of injecting nearly all its products old and new with \u201cCopilot\u201d functionality. In flashy keynotes<\/a>, slick demos<\/a> and, now, an AI key, the company\u2019s making its AI tech prominent \u2014 and betting on this to drive demand.<\/p>\n Demand isn\u2019t a sure thing. But to be fair. a few vendors have managed to turn viral AI hits into successes. Look at OpenAI, the maker of ChatGPT, which reportedly<\/a> topped $1.6 billion in annualized revenue toward the end of 2023. Generative art platform Midjourney is apparently profitable, also \u2014 and hasn\u2019t yet taken a dime of outside capital.<\/p>\n Emphasis on\u00a0a few,\u00a0<\/em>though. Most vendors, weighed down by the costs of training and running cutting-edge AI models, have had to seek larger and larger tranches of capital to stay afloat. Case in point, Anthropic is said to be raising<\/a> $750 million in a round that would bring its total raised to more than $8 billion.<\/p>\n Microsoft, together with its chip partners AMD and Intel, hopes that AI processing will increasingly move from expensive datacenters to local silicon, commoditizing AI in the process \u2014 and it might well right. Intel\u2019s new lineup of consumer chips pack custom-designed cores for running AI. Plus, new datacenter chips<\/a> like Microsoft\u2019s own could make model training a less expensive endeavor than it is currently.<\/p>\n But there\u2019s no guarantee. The real test will be seeing whether Windows users and enterprise customers, bombarded with what amounts to Copilot advertising, show an appetite for the tech \u2014 and shell out for it. If they don\u2019t, it might not be long before Microsoft has to redesign the Windows keyboard once again.<\/p>\n Here are some other AI stories of note from the past few days:<\/p>\n You may remember various examples of interesting work over the last year involving making minor changes to images<\/a> that cause machine learning models to mistake, for instance, a picture of a dog for a picture of a car. They do this by adding \u201cperturbations,\u201d minor changes to the pixels of the image, in a pattern that only the model can perceive. Or at least they thought<\/em> only the model could perceive it.<\/p>\n An experiment by Google DeepMind researchers<\/a> showed that when a picture of flowers was perturbed to appear more catlike to AI, people were more likely to describe that image as more catlike despite its definitely not looking any more like a cat. Same for other common objects like trucks and chairs.<\/p>\n Image Credits:<\/strong> Google DeepMind<\/p>\n<\/div>\n Why? How? The researchers don\u2019t really know, and the participants all felt like they were just choosing randomly (indeed the influence is, while reliable, scarcely above chance). It seems we\u2019re just more perceptive than we think \u2014 but this also has implications on safety and other measures, since it suggests that subliminal signals could indeed propagate through imagery without anyone noticing.<\/p>\n Another interesting experiment involving human perception came out of MIT this week, which used machine learning to help elucidate a particular system of language understanding<\/a>. Basically some simple sentences, like \u201cI walked to the beach,\u201d barely take any brain power to decode, while complex or confusing ones like \u201cin whose aristocratic system it effects a dismal revolution\u201d produce more and broader activation, as measured by fMRI.<\/p>\n The team compared the activation readings of humans reading a variety of such sentences with how the same sentences activated the equivalent of cortical areas in a large language model. Then they made a second model that learned how the two activation patterns corresponded to one another. This model was able to predict for novel sentences whether they would be taxing on human cognition or not. It may sound a bit arcane, but it is definitely super interesting, trust me.<\/p>\n Whether machine learning can imitate human cognition in more complex areas, like interacting with computer interfaces, is still very much an open question. There\u2019s lots of research, though, and it\u2019s always worth taking a look at. This week we have SeeAct<\/a>, a system from Ohio State researchers that works by laboriously grounding a LLM\u2019s interpretations of possible actions in real-world examples.<\/p>\n Image Credits:<\/strong> Ohio State University<\/p>\n<\/div>\n Basically you can ask a system like GPT-4V to create a reservation on a site, and it will get what its task is and that it needs to click the \u201cmake reservation\u201d button, but it doesn\u2019t really know how to do that. By improving how it perceives interfaces with explicit labels and world knowledge, it can do lots better, even if it still only succeeds a fraction of the time. These agent models have a long way to go, but expect a lot of big claims this year anyway! I just heard some today.<\/p>\n Next, check out this interesting solution to a problem I had no idea existed but which makes perfect sense. Autonomous ships are a promising area of automation, but when the sea is angry it is difficult to make sure they\u2019re on track. GPS and gyros don\u2019t cut it, and visibility can be poor too \u2014 but more importantly, the systems governing them aren\u2019t too sophisticated. So they can go wildly off target or waste fuel going on large detours if they don\u2019t know any better, a big problem if you\u2019re on battery power. I never even thought about that!<\/p>\n Korea\u2019s Maritime and Ocean University<\/a> (another thing I learned about today) proposes a more powerful pathfinding model built on simulating ship movements in a computational fluid dynamics model. They propose that this better understanding of wave action and its effect on hulls and propulsion could seriously improve the efficiency and safety of autonomous marine transport. It might even make sense to use in human-guided vessels whose captains aren\u2019t quite sure what the best angle of attack is for a given squall or wave form!<\/p>\n Last, if you want a good recap of last year\u2019s big advances in computer science, which in 2023 overlapped massively with ML research, check out Quanta\u2019s excellent review<\/a>.<\/p>\n<\/p><\/div>\n\n
More machine learnings<\/h2>\n
![\"\"](\"https:\/\/techcrunch.com\/wp-content\/uploads\/2024\/01\/action_grounding.png\")
<\/p>\n