Ashu Garg: “The challenges to scaling broadly fall into three interrelated categories: data, compute, and the limitations of next-token prediction. Each represents a barrier to progress through more data, compute, and parameters alone…Two years ago, ChatGPT emerged seemingly from nowhere, transforming our understanding of what AI can do. The next step change may be equally unexpected, arising not from raw computing power, but from a deeper understanding of the intelligence we’re trying to create.”
WSJ: “Transformers work by figuring out how every single piece of information the system takes in relates to every other piece of information it’s been fed, says Tim Dettmers, an AI research scientist at the nonprofit Allen Institute for Artificial Intelligence. That level of contextual understanding enables transformer-based AI systems to not only recognize patterns, but predict what could plausibly come next—and thus generate their own new information. And that ability can extend to data other than words. “In a sense, the models are discovering the latent structure of the data,” says Alexander Rives, chief scientist of EvolutionaryScale, which he co-founded last year after working on AI for Meta Platforms, the parent company of Facebook.”
ET: “Affluent households in urban India rose in 2024 while lower middle-class households saw a decline compared to five years ago, according to a report, underscoring a widening gap between the socio-economic class (SEC) in cities and presenting a dilemma for fast-moving consumer goods companies. The number of affluent households surged 86% while lower and lower middle-class households fell 25% this calendar year compared to 2019, according to Kantar’s ‘India at Crossroads’ report. Consequently, SEC C, D and E, mostly blue-collar people, comprised 35% of all households in urban India, declining from 52% in 2019. This shows that the urban middle class, once the cornerstone of the consumer goods market, has become the smallest cohort…Kantar pointed out that real incomes have been under pressure, leading to stagnating consumer confidence, and a sharper slowdown in consumption in the urban areas than in rural areas.”
NYTimes: “Meet FRED, a 33-year-old data tool from St. Louis, Mo., and the economics world’s most unlikely celebrity. Even if you have not interacted with FRED yourself, there is a good chance you’ve encountered him without knowing it. The tool’s signature baby blue graphs dot social media and crop up on many of the world’s most popular news websites…Many people feel that way about FRED. The website had nearly 15 million users last year, and it is on track for even more in 2024, up from fewer than 400,000 as recently as 2009. Their reasons for clicking are diverse: FRED users are coming for freshly released unemployment data, to check in on egg inflation or to find out whether business is booming in Memphis.”
Mint on Himachal Pradesh: “The state’s financial position is such that it is struggling to manage its day-to-day operations. Salaries have been delayed. Most contractual payments have been held up…The state’s expenses far outstrip its revenue. Committed expenditure, those on salaries, pensions and interest, alone accounts for 67% of its revenue expenses. That has left very little space for capital and other developmental expenses. In 2023-24, the state posted a revenue deficit of 2.6% and a fiscal deficit of 5.9%, way above the target fixed for the states.”
