There are two big topics swirling around Silicon Valley right now: Layoffs and ChatGPT. It turns out, they’re related.

Living in Silicon Valley, it’s hard not to feel like layoffs are all around me.

My next-door neighbor, a former Twitter employee, lost her job late last year. A close friend who worked at Google also got laid off, one of 12,000 in Alphabet’s recent round of pink slips. Indeed, the headlines keep coming: 11,000 jobs cut at Meta; 8,000 at Salesforce; 18,000 at Amazon.

Most tech company CEOs have blamed the layoffs on their pandemic-era hiring sprees, saying they take “full responsibility” for the decisions that led to them. (This begs the question: Is it not obvious that the responsibility for both hiring and firing decisions always lies with the CEO?) Not surprisingly, most leaders are also saying they are cutting with care—not across the board but in areas they have deemed to be too “fat” or non-essential to their company’s growth strategy. (Sadly, early data shows that one of the areas that’s being disproportionately impacted by layoffs are diversity, equity, and inclusion roles.)

But just as telling as where cuts are being made is where CEOs don’t appear to be making cuts.

“It’s important to note that while we are eliminating roles in some areas, we will continue to hire in key strategic areas,” Microsoft CEO Satya Nadella wrote in a memo to employees on Jan. 18, in which he announced that the company would be shedding 10,000 jobs. “We will continue to invest in strategic areas for our future, meaning we are allocating both our capital and talent to areas of secular growth and long-term competitiveness for the company, while divesting in other areas.”

What are those strategic areas that represent long-term competitiveness? I can think of one: artificial intelligence. Indeed, early on in his memo, Nadella says that “The next major wave of computing is being born with advances in A.I.”

But that shift in strategy is not only happening at Microsoft. Just two days after Nadella’s layoffs memo was pushed to employees and published online, Alphabet CEO Sundar Pichai sent a similar letter to Googlers. (Yes, the two companies seem to compete even when it comes to the timing of job cuts.)

“I am confident about the huge opportunity in front of us thanks to the strength of our mission, the value of our products and services, and our early investments in A.I.,” Pichai told employees. “To fully capture it, we’ll need to make tough choices.”

Those tough decisions included laying off 6% of Alphabet’s workforce, and shuttering several experimental projects. A.I. has not been completely spared from the belt-tightening—Alphabet’s DeepMind A.I. subsidiary recently shut down an outpost in Edmonton, Canada. And Insider reported that Microsoft’s Cloud+AI group, led by EVP Scott Guthrie, was impacted by the recent layoffs (though how severely, remains unclear).

Compared to other groups getting “right-sized” right now, however, A.I. is far less expendable. Few Big Tech CEOs would refer to their A.I. efforts as experimental—existential is probably a more apt word to use here, given the tense and high-stakes race brewing between Google’s late-to-the-game Bard chatbot and Microsoft’s bet on ChatGPT, the OpenAI-developed tool that’s reportedly growing faster than TikTok.

Speaking of ChatGPT, there’s no bigger signal of how priorities are shifting than the timing of Microsoft’s multi-year, multi-billion-dollar investment in its maker, OpenAI, which was officially announced just days after the company disclosed its pending job cuts.

Make no mistake: Despite the excitement over what ChatGPT and its competitors might someday enable—including, perhaps, new types of jobs—the current wave of layoffs is tough on workers, on their families, and even on the morale of employees who weren’t directly impacted. More than 100,000 people have lost their tech company jobs since the start of this year alone, according to online tracker Layoffs.fyi. Those job cuts took place across 344 different companies, ranging from the largest tech titans to small-scale startups. One company that’s not on the massive (and massively depressing) database compiled by Layoffs.fyi? No surprise here: OpenAI.

With that, here’s the rest of this week’s A.I. news.

Michal Lev-Ram

michal.levram@fortune.com

A.I. IN THE NEWS

Google's Bard fallout continues. Alphabet's now infamous error during its demonstration of Bard, its rival to Bing's new OpenAI search tool, didn't just take a toll on its stock price. The fallout from the much-mocked incident is now being felt inside the company too. According to a recent piece published by CNBC, Google employees took to an internal forum to complain about the company’s recent announcement of Bard, saying the effort was “rushed,” “botched,” and “comically short-sighted.” The search giant has its hands full at the moment, as employees grapple with a recent round of layoffs (see above) and a mandate to refocus on growth areas like A.I.—not to mention the need to show that it can not only compete but lead in an area it has long dominated.

So, about that Bing demo... Meanwhile, Bing's new A.I.-powered search is also getting a bit of renewed scrutiny. In a recent blog post, Dmitri Brereton—who describes himself as focused on "independent research on search engines and AI"—looked back at last week's splashy launch event and found a slew of factual errors that were overlooked amid the hoopla. Some of the most frequent mistakes involved the financial data that the A.I. cited in its summaries of earnings reports. In some cases, the A.I. conflated metrics (adjusted vs non-adjusted gross margins); in other cases, the A.I. cited data that appeared to be completely invented.

But at least their handwriting is easier to read? It turns out that A.I. algorithms aren’t that effective at diagnosing Covid-19 from the sound of a person’s cough. The ambitious project, started early on in the pandemic, involved a variety of startups, researchers, and academic institutions. But a recent independent review led by the Alan Turing Institute and Royal Statistical Society in the U.K. found that cough-analyzing algorithms performed worse than user-reported systems and demographic data. So much for using audio-based A.I. as a Covid-19 screening tool. But at least we have cheap tests now!

FORTUNE ON A.I.

Eric Schmidt says A.I. could be the military’s new nuke—but only if the Pentagon acts like a tech firm—by Tristan Bove

Amazon’s self-driving taxis hit the road in California, shuttling employees to and from work—by Chris Morris

ChatGPT creator: A.I. will be a ‘force for economic empowerment’ and not a job killer—plus make a lot of people rich—by Christiaan Hetzner

McDonald’s new A.I. ordering system isn’t going exactly to plan as bewildered TikTokers document drive-thru fails: ‘Not a ketchup packet! Oh, my God’—by Eleanor Pringle

BuzzFeed is rolling out A.I. quizzes that can write breakup messages or create movie ideas—but humans are still involved—by Prarthan Prakash

BRAIN FOOD

Talking to Machines At this point, I think we all know that generative A.I. is eating the world—or will soon enough. But just like regular old software upended the way we do everything, how will A.I. change the way we communicate, consume, and collaborate? That last word is key, because as this recent article from The Atlantic tells us, A.I. could "become the overlay for not only search engines, but also creative work, busywork, memo writing, research, homework, sketching, outlining, storyboarding, and teaching."

Indeed, all of that work involves both machines and humans. And, as this story rightfully asserts, that kind of collaboration would require people to learn to effectively chat with, well, chatbots in ways most of us haven’t done before. Case in point: We’re all used to asking Siri or Alexa to look up a location, play us a song, or even to tell us a joke. But what if we were to write that joke with a chatbot, in order to come up with something truly new and novel? That would take not just querying but actually collaborating with an A.I. system—in other words, talking to it. This is not a skill that’s natural to most of us. But the need for us to learn how to effectively communicate with machines in order to create something, not just to access already existing information, is sure to grow, given the billions of dollars currently being thrown at systems like ChatGPT.

Naturally, there’s an app for that, or rather, a growing “cottage industry” of those who already know how to speak to machines. As the article points out, PromptBase is a marketplace where you can “purchase a few lines of text to feed into any number of generative-AI models” in order to achieve optimal results. And individuals are already hawking their A.I. communication skills to whoever will listen—and pay. (One solid piece of advice: knowing the model you’re speaking to, since not all A.I. systems are created equal.) Of course, some of us are challenged even when talking to other mere humans. But the way things are progressing, you might be better off learning how to effectively talk to machines.