Lots of ChatGPT users have been complaining that the powerful AI model has been getting “lazier” and “dumber.” When I put this to ChatGPT, it reminded me that as an AI language model, it doesn’t have feelings or consciousness. Only humans can get dumber.
The more technical question then is, is it getting less accurate? An executive at OpenAI initially said the users complaining about degraded performance were wrong.
Then research out of Stanford University and UC Berkeley found that the performance of ChatGPT’s 3.5 version and 4.0 version varied greatly, with the newer model underperforming on some tasks. For example, ChatGPT turned out “more formatting mistakes in code generation” last month than it did earlier in the year, according to the research.
This has important implications. The promise of machine learning is that, well, it learns. But as the research paper noted, it is now an “interesting question whether an LLM service like GPT4 is consistently getting ‘better’ over time.”
It’s hard to buy a house right now. Those who already own a home aren’t selling, for fear of replacing their current low mortgage rate with something much higher. Prices aren’t budging. And competition remains fierce.
Wannabe homeowners might be best placed looking at a brand-spanking new build. Recent data shows a surge in construction, and many builders are dangling perks like discounts and lower rates.
BlackRock; Joel Saget/AFP via Getty Images; Alyssa Powell/Insider
It’s a strange time for BlackRock. Larry Fink has spent half his life running the firm, and the asset manager’s board has been busy sizing up possible replacements for him.
However, casting a shadow over Fink’s legacy and eventual departure is the backlash over BlackRock’s environmental, social, and governance investing. It’s already made the firm and Fink a favorite political target for Republicans — now the scrutiny is adding a new layer of complexity to BlackRock’s succession plans.
Jahi Chikwendiu/The Washington Post via Getty Images
Data-center development is booming across rural America. The industry had long been clustered in a handful of well-established markets like Chicago and Silicon Valley — but now, as AI stokes demand for data centers, developers are expanding to unexpected places.
That’s because AI requires massive amounts of energy — and the industry’s core markets simply don’t have enough of it. Instead, developers are now building massive data centers in rural towns, where they can tap pockets of electricity on the grid.
We all know that Google pays well. But an internal spreadsheet gives us our best look yet at how much exactly the tech giant is paying staff. We’ve obtained the leaked data and built a table of searchable roles with information on salaries, equity, and bonuses.
The best-paid are software engineers, who could make over $1 million yearly in total compensation.
“By combining a Neuralink implant and a robotic arm or leg, we believe we can give amputees basically a cyborg body that is incredibly capable. The $6 million man in real life. But it won’t cost $6 million — a $60,000 man.”
— Elon Musk speaking on Tesla’s latest earnings call, where he touched on everything from “Game of Thrones” and Nvidia to Warren Buffett and the economy.