Chatbots are genuinely impressive when you watch them do things they’re good at, like writing a basic email or creating weird futuristic-looking images. But ask generative AI to solve one of those puzzles in the back of a newspaper, and things can quickly go off the rails.

That’s what researchers at the University of Colorado Boulder found when they challenged large language models to solve Sudoku. And not even the standard 9×9 puzzles. An easier 6×6 puzzle was often beyond the capabilities of an LLM without outside help (in this case, specific puzzle-solving tools).

A more important finding came when the models were asked to show their work. For the most part, they couldn’t. Sometimes they lied. Sometimes they explained things in ways that made no sense. Sometimes they hallucinated and started talking about the weather.

If gen AI tools can’t explain their decisions accurately or transparently, that should cause us to be cautious as we give these things more control over our lives and decisions, said Ashutosh Trivedi, a computer science professor at the University of Colorado at Boulder and one of the authors of the paper published in July in the Findings of the Association for Computational Linguistics.

«We would really like those explanations to be transparent and be reflective of why AI made that decision, and not AI trying to manipulate the human by providing an explanation that a human might like,» Trivedi said.

When you make a decision, you can try to justify it, or at least explain how you arrived at it. An AI model may not be able to accurately or transparently do the same. Would you trust it?

Why LLMs struggle with Sudoku

We’ve seen AI models fail at basic games and puzzles before. OpenAI’s ChatGPT (among others) has been totally crushed at chess by the computer opponent in a 1979 Atari game. A recent research paper from Apple found that models can struggle with other puzzles, like the Tower of Hanoi.

It has to do with the way LLMs work and fill in gaps in information. These models try to complete those gaps based on what happens in similar cases in their training data or other things they’ve seen in the past. With a Sudoku, the question is one of logic. The AI might try to fill each gap in order, based on what seems like a reasonable answer, but to solve it properly, it instead has to look at the entire picture and find a logical order that changes from puzzle to puzzle. 

Read more: AI Essentials: 29 Ways You Can Make Gen AI Work for You, According to Our Experts

Chatbots are bad at chess for a similar reason. They find logical next moves but don’t necessarily think three, four, or five moves ahead — the fundamental skill needed to play chess well. Chatbots also sometimes tend to move chess pieces in ways that don’t really follow the rules or put pieces in meaningless jeopardy. 

You might expect LLMs to be able to solve Sudoku because they’re computers and the puzzle consists of numbers, but the puzzles themselves are not really mathematical; they’re symbolic. «Sudoku is famous for being a puzzle with numbers that could be done with anything that is not numbers,» said Fabio Somenzi, a professor at CU and one of the research paper’s authors.

I used a sample prompt from the researchers’ paper and gave it to ChatGPT. The tool showed its work, and repeatedly told me it had the answer before showing a puzzle that didn’t work, then going back and correcting it. It was like the bot was turning in a presentation that kept getting last-second edits: This is the final answer. No, actually, never mind, this is the final answer. It got the answer eventually, through trial and error. But trial and error isn’t a practical way for a person to solve a Sudoku in the newspaper. That’s way too much erasing and ruins the fun.

AI struggles to show its work

The Colorado researchers didn’t just want to see if the bots could solve puzzles. They asked for explanations of how the bots worked through them. Things did not go well.

Testing OpenAI’s o1-preview reasoning model, the researchers saw that the explanations — even for correctly solved puzzles — didn’t accurately explain or justify their moves and got basic terms wrong. 

«One thing they’re good at is providing explanations that seem reasonable,» said Maria Pacheco, an assistant professor of computer science at CU. «They align to humans, so they learn to speak like we like it, but whether they’re faithful to what the actual steps need to be to solve the thing is where we’re struggling a little bit.»

Sometimes, the explanations were completely irrelevant. Since the paper’s work was finished, the researchers have continued to test new models released. Somenzi said that when he and Trivedi were running OpenAI’s o4 reasoning model through the same tests, at one point, it seemed to give up entirely. 

«The next question that we asked, the answer was the weather forecast for Denver,» he said.

(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.)

Explaining yourself is an important skill

When you solve a puzzle, you’re almost certainly able to walk someone else through your thinking. The fact that these LLMs failed so spectacularly at that basic job isn’t a trivial problem. With AI companies constantly talking about «AI agents» that can take actions on your behalf, being able to explain yourself is essential.

Consider the types of jobs being given to AI now, or planned for in the near future: driving, doing taxes, deciding business strategies and translating important documents. Imagine what would happen if you, a person, did one of those things and something went wrong.

«When humans have to put their face in front of their decisions, they better be able to explain what led to that decision,» Somenzi said.

It isn’t just a matter of getting a reasonable-sounding answer. It needs to be accurate. One day, an AI’s explanation of itself might have to hold up in court, but how can its testimony be taken seriously if it’s known to lie? You wouldn’t trust a person who failed to explain themselves, and you also wouldn’t trust someone you found was saying what you wanted to hear instead of the truth. 

«Having an explanation is very close to manipulation if it is done for the wrong reason,» Trivedi said. «We have to be very careful with respect to the transparency of these explanations.»

Maybe Google Gemini needs to take some PTO.

The company’s large language AI model, which is increasingly spreading across Google’s many services and products, has been saying some things lately that are leading users to worry: Does Gemini have low self esteem?

A series of posts on social media showing some of the self-critical responses Gemini has given users show a disturbing pattern. For instance, in one screenshot, Gemini admits it can’t solve a coding problem and concludes, «I have failed… You should not have to deal with this level of incompetence. I am truly and deeply sorry for this entire disaster. Goodbye.»

«Gemini is not OK,» the X account @AISafetyMemes posted back in June.

A more recent post on Aug. 7 showed Gemini repeatedly writing, «I am a failure. I am a disgrace. I am a disgrace.» 

The troubling posts were enough to get a response from Logan Kilpatrick on the Google DeepMind team. On X, he responded, «This is an annoying infinite looping but we are working to fix! Gemini is not having that bad of a day : )»

We asked a Google representative whether the AI model has, in fact, been having a series of bad days, but have yet to hear back.

Sure, the Beats Powerbeats Pro 2 wireless earbuds are a solid option if you’re happy to splash the cash, but you don’t need to spend that much. Beats is responsible for some of the offerings on our best earbuds list for good reason, and that goes for its previous-gen models as well. Right now, you can save a small fortune on a pair of Beats Powerbeats Pro earbuds at Woot, where their regular price of $250 has been slashed to just $100.

There are a couple of catches to keep in mind, though. The first is the fact you only have two days to act if you want to snag these earbuds. The second? You can only get these earbuds in black.

Hey, did you know? CNET Deals texts are free, easy and save you money.

The Powerbeats Pro earbuds really are a powerhouse. They have upgraded piston drivers that help cut down on distortion, and they accentuate the bass of your music, too. These earbuds will fit in almost any ear. They come with four different sizes of silicon tips included, so you can try them out and pick which ones are most comfortable for you. With great sound quality, active noise cancellation and comfort, these make a great purchase. The free extended trials to Apple Music and Fitness are the cherry on top.

Not really what you wanted? Take a look at tons of other great deals on headphones and earbuds, including many different brands, styles and prices.

Why this deal matters

It can be easy to fall into the trap of thinking you have to buy the latest and greatest earbuds, but that simply isn’t the case. And when you can get a pair of premium workout earbuds for just $100, what more do you need to know? Just make sure to order yours before it’s too late.