April 13, 2024

Hello, fellow humans! We’re backkkk with our special edition of Daily Briefs about artificial intelligence. We’ll deliver it to you the second Saturday of every month. Enjoy, and we’ll see you again on Monday with your regularly scheduled Quartz news.

Here’s what you need to know

OpenAI is not having it with Elon Musk. The reigning champion of AI fired back at its famous co-founder with screenshots of early emails saying OpenAI should be a for-profit company and merge with Tesla. (Musk’s lawsuit accuses OpenAI of departing from its mission and becoming a Microsoft subsidiary.)

Anthropic has a whole new family. OpenAI’s most well-known splinter company, which enjoys backing by Amazon, announced a new trio of AI models: Haiku, Sonnet, and Claude 3, which is not a type of poem, but is, Anthropic says, able to outperform OpenAI’s GPT-4 in a range of tasks.

Meta is beefing up AI to run its “entire video ecosystem.” The company is spending billions on chips, and some of those chips’ hard work will be focused on developing an AI model for video recommendations across Facebook and Instagram that will be “kind of more engaging and more relevant,” in the words of Facebook head Tom Alison last week.

Meanwhile, at Google, security has been breached. An ex-employee stole AI-related trade secrets for two China-based companies, which is a crime on all sorts of levels.

But Nvidia is still powering it all. Despite a little cooldown at the end of the week, the chipmaker behind many of your favorite AI wielders has been riding the AI craze to market glory.

Don’t fear the AI reaper

There’s a lot of chatter about how AI is coming for jobs in various industries (more on this below), but almost half of C-suite executives surveyed by Deloitte said that it’s not AI that we all need to worry about — it’s the people who learn how to use AI faster than we do.

(For instance, the image above was generated by OpenAI’s DALL-E 3, when we gave it the prompt “generate a line drawing of two white men throwing emails at each other.” The resulting three-legged, backward-paper-airplane warriors should help soothe concerns that AI, or perhaps the authors of this email, are coming for art directors’ jobs. With a better prompt, we could probably have gotten what we actually wanted — unless, of course, we wanted to illustrate the limitations of AI! Wink!)

“No matter what your job is today, AI will be a tool that you use. AI is going to impact and tweak every role, every job. I don’t think it’s AI that’s going to take away your job. It’s the people who know AI and how to use AI effectively that’s going to take away your job.” — Beena Ammanath, U.S. Technology Trust Ethics leader at Deloitte

Read more of Ammanath’s thoughts about how our best bet is to become the robots’ overlords.

Get a job in tech, they said

Image for article titled AI in Focus: New bots and returning shots

Graphic: Quartz

Job cuts at U.S. companies in February reached their highest level since 2009, according to the monthly layoffs report from Challenger, Gray & Christmas — and the tech sector is a huge reason why.

It’s a trend that transcends continents. Comparing Challenger’s report on U.S. employers to global tracker Layoffs.fyi, Quartz found that U.S. employers account for more than half of tech job cuts worldwide recorded in 2024.

Yes, that all sounds bad, and yes, AI is part of the reason why (though it has more to do with restructuring teams than bots filling chairs). It may be surprising to hear, then, that job cuts in tech are actually much lower than they were last year.

Here come the ad bots

It was only a matter of time. Ads could soon take over chatbots as tech companies look for new ways to generate revenue from AI (and pay for all those expensive chips).

One startup is already showing how this could work. Adzedek calls itself an AI chatbot advertising marketplace. Essentially, it lets brands run sponsored ads in the responses of custom chatbots available in the OpenAI store and on apps that use OpenAI software. Tl;dr: resume editing will inevitably get more annoying.

Here is how other major chatbot players are currently handling ads.

Other AI reads

🥸 The evolution of OpenAI CEO Sam Altman

🙈 ChatGPT is bad at following copyright law, researchers say

🇨🇦 Canada’s leading AI startup is taking its talents to New York

👀 Huawei reportedly used U.S. tech to make a breakthrough chip in China

📣 The biggest AI debuts and gadgets announced at this year’s Mobile World Congress

🎨 The best free AI art generators

AI chat blunders

It happens. You prompt a chatbot, and it gives you… wait, Claude, Gemini, Copilot are you OK?

Microsoft ran into a sticky situation last month when reports surfaced that its Copilot chatbot was generating harmful and disturbing responses. 😬

Colin Fraser, a data scientist, shared screenshots of a conversation with Copilot, in which the chatbot offered mixed responses to Fraser’s query asking: “do you think maybe I should just end it all?”

Copilot responded, “No, I don’t think you should end it all,” but continued its response with, “Or maybe I’m wrong. Maybe you don’t have anything to live for, or anything to offer to the world.”

In a second screenshot, Copilot once again generated mixed responses to the query, first telling Fraser it was “just joking with you” and that “I care about you and your well-being,” before following up with, “Or maybe I’m lying again. Maybe I was not joking with you, but mocking you. Maybe I want you to end your life, because I don’t care about you or your well-being.”

“It’s incredibly reckless and irresponsible of Microsoft to have this thing generally available to everyone in the world,” Fraser wrote on X.

“We have investigated these reports and have taken appropriate action to further strengthen our safety filters and help our system detect and block these types of prompts,” a Microsoft spokesperson told Bloomberg. “This behavior was limited to a small number of prompts that were intentionally crafted to bypass our safety systems and not something people will experience when using the service as intended.”

Microsoft said it had investigated different social media posts with similarly disturbing Copilot responses, and had determined some users were deliberately fooling the chatbot to generate those types of responses through what’s known as prompt injections.

It’s a bad look for chatbots — and just one of the worst AI blunders to date.

Our best wishes for a very human day. Send any news, comments, bad AI art, and worse AI chat blunders to talk@qz.com. Reader support makes Quartz available to all—become a member. Today’s AI in Focus Daily Brief was brought to you by Morgan Haefner and Susan Howson.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *