Technologies

Computing Guru Criticizes ChatGPT AI Tech for Making Things Up

Vint Cerf, who helped create the internet’s core network technology, hopes engineers can improve artificial intelligence’s shortcoming.

Vint Cerf, one of the founding fathers of the internet, has some harsh words for the suddenly hot technology behind the ChatGPT AI chatbot: «Snake oil.»

Google’s internet evangelist wasn’t completely down on the artificial intelligence technology behind ChatGPT and Google’s own competing Bard, called a large language model. But, speaking Monday at Celesta Capital’s TechSurge Summit, he did warn about ethical issues of a technology that can generate plausible sounding but incorrect information even when trained on a foundation of factual material.

If an executive tried to get him to apply ChatGPT to some business problem, his response would be to call it snake oil, referring to bogus medicines that quacks sold in the 1800s, he said. Another ChatGPT metaphor involved kitchen appliances.

«It’s like a salad shooter — you know how the lettuce goes all over everywhere,» Cerf said. «The facts are all over everywhere, and it mixes them together because it doesn’t know any better.»

Cerf shared the 2004 Turing Award, the top prize in computing, for helping to develop the internet foundation called TCP/IP, which shuttles data from one computer to another by breaking it into small, individually addressed packets that can take different routes from source to destination. He’s not an AI researcher, but he’s a computing engineer who’d like to see his colleagues improve AI’s shortcomings.

OpenAI’s ChatGPT and competitors like Google’s Bard hold the potential to significantly transform our online lives by answering questions, drafting emails, summarizing presentations and performing many other tasks. Microsoft has begun building OpenAI’s language technology into its Bing search engine in a significant challenge to Google, but it uses its own index of the web to try to «ground» OpenAI’s flights of fancy with authoritative, trustworthy documents.

Cerf said he was surprised to learn that ChatGPT could fabricate bogus information from a factual foundation. «I asked it, ‘Write me a biography of Vint Cerf.’ It got a bunch of things wrong,» Cerf said. That’s when he learned the technology’s inner workings — that it uses statistical patterns spotted from huge amounts of training data to construct its response.

«It knows how to string a sentence together that’s grammatically likely to be correct,» but it has no true knowledge of what it’s saying, Cerf said. «We are a long way away from the self-awareness we want.»

OpenAI, which earlier in February launched a $20 per month plan to use ChatGPT, has been clear about about the technology’s shortcomings but aims to improve it through «continuous iteration.»

«ChatGPT sometimes writes plausible-sounding but incorrect or nonsensical answers. Fixing this issue is challenging,» the AI research lab said when it launched ChatGPT in November.

Cerf hopes for progress, too. «Engineers like me should be responsible for trying to find a way to tame some of these technologies so they are less likely to cause trouble,» he said.

Cerf’s comments stood in contrast to those of another Turing award winner at the conference, chip design pioneer and former Stanford President John Hennessy, who offered a more optimistic assessment of AI.

Editors’ note: CNET is using an AI engine to create some personalfinance explainers that are edited and fact-checked by our editors. Formore, see this post.

Technologies

OpenAI Will Loosen ChatGPT’s Mental Health Guardrails and Allow Erotica for Adult Users

Sam Altman said the company will ease limits for adults after rolling out age verification.

ChatGPT is treading cautiously right now, but the chatbot may become more risqué by the end of the year.

In recent weeks, the generative AI chatbot has been operating under somewhat stringent limitations, as OpenAI tried to address concerns that it was not handling sensitive mental health issues well. But CEO Sam Altman said in a post on X Tuesday that the company would ease some of those restrictions because it’s «been able to mitigate the serious mental health issues.»

Though Altman didn’t elaborate on what tools are being used to address the problem, OpenAI recently announced new parental controls in ChatGPT.

CNET reached out to OpenAI for details, but the company did not immediately respond to a request for comment. (Disclosure: Ziff Davis, CNET’s parent company, in April filed a lawsuit against OpenAI, alleging it infringed Ziff Davis copyrights in training and operating its AI systems.)

Other changes are also expected. Altman said the company could allow «erotica» for verified adult users as it implements an «age-gating» system, or age-restricted content, in December. The mature content is part of the company’s «treat adult users like adults» principle, Altman said.

Altman’s post also announced a new version of ChatGPT in the next few weeks, with a personality that behaves more like the company’s GPT-4o model. Chatbot users had complained after the company replaced 4o with the impersonal GPT-5 earlier this year, saying the new version lacked the engaging and fun personality of previous chatbot models.

«If you want your ChatGPT to respond in a very human-like way, or use a ton of emoji, or act like a friend, ChatGPT should do it (but only if you want it, not because we are usage-maxxing),» Altman wrote.

Don’t miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.

After OpenAI was sued by parents who alleged ChatGPT contributed to their teen son’s suicide, the company imposed an array of new restrictions and changes, including parental controls, alerts for risky behavior and a teen-friendly version of the chatbot. In the summer, OpenAI implemented break reminders that encourage people to occasionally stop chatting with the bot.

On Tuesday, the company also announced the creation of a council of experts on AI and well-being, including some with expertise in psychology and human behavior.

This comes as lawmakers and regulators are ringing the alarm on the risks AI tools pose to people, especially children. On Monday, California Governor Gavin Newsom signed new restrictions on AI companion chatbots into law. Last month, the Federal Trade Commission launched an investigation into several AI companies, including OpenAI.

Technologies

Today’s NYT Strands Hints, Answers and Help for Oct. 15 #591

Here are hints — and the answers — for the NYT Strands puzzle for Oct. 15, No. 591.

Looking for the most recent Strands answer? Click here for our daily Strands hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle, Connections and Connections: Sports Edition puzzles.

Today’s NYT Strands puzzle is a fun one, once you understand the theme. Some of the answers are a bit tough to unscramble, so if you need hints and answers, read on.

I go into depth about the rules for Strands in this story.

If you’re looking for today’s Wordle, Connections and Mini Crossword answers, you can visit CNET’s NYT puzzle hints page.

Read more: NYT Connections Turns 1: These Are the 5 Toughest Puzzles So Far

Hint for today’s Strands puzzle

Today’s Strands theme is: Going up?

If that doesn’t help you, here’s a clue: Not an escalator, but…

Clue words to unlock in-game hints

Your goal is to find hidden words that fit the puzzle’s theme. If you’re stuck, find any words you can. Every time you find three words of four letters or more, Strands will reveal one of the theme words. These are the words I used to get those hints but any words of four or more letters that you find will work:

ROTATE, LOBE, NOPE, RATS, STAR, SAME, LOSE, VOTE, BUTTE, SAMS, BAMS

Answers for today’s Strands puzzle

These are the answers that tie into the theme. The goal of the puzzle is to find them all, including the spangram, a theme word that reaches from one side of the puzzle to the other. When you have all of them (I originally thought there were always eight but learned that the number can vary), every letter on the board will be used. Here are the nonspangram answers:

ALARM, OPEN, CLOSE, LOBBY, GROUND, BASEMENT

Today’s Strands spangram

Today’s Strands spangram is ELEVATORBUTTONS. To find it, look for the E that’s three letters to the right on the bottom row, and wind straight up, and then straight down.

Technologies

Today’s NYT Connections: Sports Edition Hints and Answers for Oct. 15, #387

Here are hints and the answers for the NYT Connections: Sports Edition puzzle for Oct. 15, No. 387.

Looking for the most recent regular Connections answers? Click here for today’s Connections hints, as well as our daily answers and hints for The New York Times Mini Crossword, Wordle and Strands puzzles.

Today’s Connections: Sports Edition is packed with abbreviations, so if that’s your thing, it’s a good day for you. If you’re struggling but still want to solve it, read on for hints and the answers.

Connections: Sports Edition is published by The Athletic, the subscription-based sports journalism site owned by the Times. It doesn’t show up in the NYT Games app but appears in The Athletic’s own app. Or you can play it for free online.

Read more: NYT Connections: Sports Edition Puzzle Comes Out of Beta

Hints for today’s Connections: Sports Edition groups

Here are four hints for the groupings in today’s Connections: Sports Edition puzzle, ranked from the easiest yellow group to the tough (and sometimes bizarre) purple group.

Yellow group hint: Hoops.

Green group hint: Are you in a league?

Blue group hint: Alma maters.

Purple group hint: Knock it out of the park.