That's Harder Than It Sounds Proposed regulations in China would mean the end of flush handles on car doors, with precious little time to roll out the changes. Car door handles seem innocuous. Tesla's electronic, retractable ones--since imitated by plenty of global automakers--have become a symbol of the automaker's willingness to work from design-first principles, reimagining what the car of the future might look like, electric-style. But in September, the National Highway Traffic Safety Administration launched an investigation into the Tesla 2021 Model Y's door handles. More than 140 consumers have complained to the National Highway Traffic Safety Administration (NHTSA) about the door handles, according to a Bloomberg report published last month.

A.I. is coming for knowledge work, and yet college seems to be getting easier. Does something need to change? Around twenty years ago, when I was a graduate student in English, I taught a class in a special observation room at my university's teaching center. My students and I sat around a long oval table while cameras recorded us. I can't remember which novel we discussed, but I do know what I learned when I watched the tape afterward, with a teaching coach. She pointed out that, when I was calling on students, I often looked to my right, missing the raised hands on my left. I didn't let silences go on long enough, instead speaking just when a student had worked up the courage to talk. On the plus side, she noticed I'd been using a technique she liked, which I'd borrowed from a professor of mine: it was like cold-calling, except that, after you'd surprised a student with a challenging question, you told them that you'd circle back in a few minutes, to give them time to consider what they'd say.

AI slop is flooding every single digital platform, and music streaming services are no exception--so much so, even someone who generally avoids AI might find themselves unknowingly listening to a robot hornily singing about butts. Take the sordid saga of "Make Love to My Shitter," an AI-generated track from an artist called BannedVinylCollection. Brace Belden, a host of the popular politics podcast TrueAnon, says that Spotify recently queued up the bawdy song after he'd finished listening to alt-country legend Lucinda Williams' 1992 album Sweet Old World. "I didn't realize the song was AI at first," he says. "I thought it might've been some obscene joke record from the 80s or 90s."

We study online classification under smoothed adversaries. In this setting, at each time point, the adversary draws an example from a distribution that has a bounded density with respect to a fixed base measure, which is known apriori to the learner. For binary classification and scalar-valued regression, previous works [Haghtalab et al., 2020, Block et al., 2022] have shown that smoothed online learning is as easy as learning in the iid batch setting under PAC model. However, we show that smoothed online classification can be harder than the iid batch classification when the label space is unbounded. In particular, we construct a hypothesis class that is learnable in the iid batch setting under the PAC model but is not learnable under the smoothed online model.

We study the problem of PAC learning \gamma -margin halfspaces with Massart noise. We propose a simple proper learning algorithm, the Perspectron, that has sample complexity \widetilde{O}((\epsilon\gamma) {-2}) and achieves classification error at most \eta \epsilon where \eta is the Massart noise rate. Prior works (DGT19, CKMY20) came with worse sample complexity guarantees (in both \epsilon and \gamma) or could only handle random classification noise (DDKWZ23,KITBMV23)--- a much milder noise assumption. We also show that our results extend to the more challenging setting of learning generalized linear models with a known link function under Massart noise, achieving a similar sample complexity to the halfspace case. This significantly improves upon the prior state-of-the-art in this setting due to CKMY20, who introduced this model.

Large language models are pre-trained on ever-growing token budgets under the assumption that better pre-training performance translates to improved downstream models. In this work, we challenge this assumption and show that extended pre-training can make models harder to fine-tune, leading to degraded final performance. We term this phenomenon catastrophic overtraining. For example, the instruction-tuned OLMo-1B model pre-trained on 3T tokens leads to over 2% worse performance on multiple standard LLM benchmarks than its 2.3T token counterpart. Through controlled experiments and theoretical analysis, we show that catastrophic overtraining arises from a systematic increase in the broad sensitivity of pre-trained parameters to modifications, including but not limited to fine-tuning. Our findings call for a critical reassessment of pre-training design that considers the downstream adaptability of the model.

Researchers and futurists have been talking for decades about the day when intelligent software agents will act as personal assistants, tutors, and advisers. Apple produced its famous Knowledge Navigator video in 1987. I seem to remember attending an MIT Media Lab event in the 1990s about software agents, where the moderator appeared as a butler, in a bowler hat. With the advent of generative AI, that gauzy vision of software as aide-de-camp has suddenly come into focus. WIRED's Will Knight provided an overview this week of what's available now and what's imminent.

Do you have trouble hearing what people are saying on TV? There are many ways you can improve the audio if you know where to look. More people have turned on closed captions in recent years, reading along as they watch a movie or show. Most say they have trouble hearing dialogue. But even clean audio can be harder to hear now that TVs have become so thin.

Phase retrieval (PR) concerns the recovery of complex phases from complex magnitudes. We identify the connection between the difficulty level and the number and variety of symmetries in PR problems. We focus on the most difficult far-field PR (FFPR), and propose a novel method using double deep image priors. In realistic evaluation, our method outperforms all competing methods by large margins. As a single-instance method, our method requires no training data and minimal hyperparameter tuning, and hence enjoys good practicality.

Over the last couple of years, the retail industry has been navigating against brutal headwinds: the worst pandemic in 100 years, global supply chain disruptions, and accelerating inflation -- all made worse by 1.1 million unfilled retail jobs. With a recession looming, retail businesses find themselves between a rock and a hard place. Still, every crisis brings about some positive changes. Retailers are discovering new strategies for customer service, supply chains, inventory management, pricing and promotion. They are preparing their brick-and-mortar stores for the digital age, reinventing legacy systems and beginning to tackle such advanced technologies as artificial intelligence (AI).