Technologies

Bing AI Bungles Search Results at Times, Just Like Google

Expect search engines to tread carefully as tests reveal problems with AI language skills.

A close reading of Bing’s AI-boosted search results reveals that the website can make the same kinds of errors that are apparent in the ChatGPT technology foundation it uses and in Google’s competing Bard.

A new version of Bing that’s in limited testing employs large language model, or LLM, technology from OpenAI, the research lab Microsoft invested in and that grabbed the tech spotlight with its ChatGPT AI chatbot. ChatGPT and related technology, trained on enormous swaths of the internet, can produce remarkable results, but they don’t truly know facts and they can make errors.

Bing tries to avoid such errors by «grounding» results with Microsoft’s search technology, assessing the veracity and authority of source documents and offering links to sources, so people can evaluate the results on their own better. But AI and search engine researcher Dmitri Brereton and others have spotted errors in Bing’s demo, including flawed financial data from a Gap quarterly earnings report.

It’s a buzzkill moment for AI. The technology truly can produce remarkable and useful results, but the trouble is assessing when it isn’t doing that. Expect more caution as the search engine industry tries to find the right formula.

Google felt the pain after a demo last week when its Bard tool, which isn’t yet publicly available, produced erroneous information about the James Webb Space Telescope.

Microsoft offered a similar response Tuesday to what Google said about its gaffe: «We recognize that there is still work to be done and are expecting that the system may make mistakes during this preview period, which is why the feedback is critical so we can learn and help the models get better.»

One basic problem is that large language models, even if trained using text like academic papers and Wikipedia entries that have passed some degree of scrutiny, don’t necessarily assemble factual responses from that raw material.

As internet pioneer and Google researcher Vint Cerf said Monday, AI is «like a salad shooter,» scattering facts all over the kitchen but not truly knowing what it’s producing. «We are a long way away from the self-awareness we want,» he said in a talk at the TechSurge Summit.

Summarizing documents might appear to be within AI’s advanced language processing abilities, but constructing human-readable sentences without drawing in inappropriate information can be difficult. For example, in this request to summarize a spec sheet for Canon’s new R8 mirrorless camera, Bing technology rattled off many features that actually are found in Canon’s earlier R5.

Another high-profile computer scientist, former Stanford University Professor John Hennessy, was more bullish than Cerf, praising AI for its language skills in understanding search queries and generating results that often are correct. But he also cautioned, «It’s always confident that it has the right answer even when it doesn’t.»

OpenAI itself offers the same warning. «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.

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

Technologies

Today’s NYT Connections Hints, Answers and Help for May 24, #713

Hints and answers for Connections for May 24, #713.

Looking for the most recent 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, Connections: Sports Edition and Strands puzzles.

Today’s Connections puzzle has a fun variety of categories. The purple one appeals to my English major heart. Read on for clues and today’s Connections answers.

The Times now has a Connections Bot, like the one for Wordle. Go there after you play to receive a numeric score and to have the program analyze your answers. Players who are registered with the Times Games section can now nerd out by following their progress, including number of puzzles completed, win rate, number of times they nabbed a perfect score and their win streak.

Read more: Hints, Tips and Strategies to Help You Win at NYT Connections Every Time

Hints for today’s Connections groups

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

Yellow group hint: Goo-goo.

Green group hint: Not shirts.

Blue group hint: City that never sleeps.

Purple group hint: Acclaimed writers.

Answers for today’s Connections groups

Yellow group: Baby gear.

Green group: Kinds of pants minus «s.»

Blue group: New York sports team members.

Purple group: Black women authors.

Read more: Wordle Cheat Sheet: Here Are the Most Popular Letters Used in English Words

What are today’s Connections answers?

The yellow words in today’s Connections

The theme is baby gear. The four answers are bib, bottle, monitor and stroller.

The green words in today’s Connections

The theme is kinds of pants minus «s.» The four answers are capri, jean, jogger and slack.

The blue words in today’s Connections

The theme is New York sports team members. The four answers are Jet, Met, Net and Ranger.

The purple words in today’s Connections

The theme is black women authors. The four answers are Butler, Gay, Hooks and Walker.

Technologies

Today’s NYT Mini Crossword Answers for Tuesday, May 20

Here are the answers for The New York Times Mini Crossword for May 20.

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

Today’s NYT Mini Crossword is a fun one, and now I’m singing the song from 1-Across in my head. Need some help with today’s Mini Crossword? Read on. And if you could use some hints and guidance for daily solving, check out our Mini Crossword tips.

The Mini Crossword is just one of many games in the Times’ games collection. If you’re looking for today’s Wordle, Connections, Connections: Sports Edition and Strands answers, you can visit CNET’s NYT puzzle hints page.

Read more: Tips and Tricks for Solving The New York Times Mini Crossword

Let’s get at those Mini Crossword clues and answers.

Mini across clues and answers

1A clue: «Pink ___ Club» (Chappell Roan hit)
Answer: PONY

5A clue: Instrument that might be made with a comb and wax paper
Answer: KAZOO

6A clue: How bedtime stories are often read
Answer: ALOUD

7A clue: On edge
Answer: TENSE

8A clue: Short Instagram video
Answer: REEL

Mini down clues and answers

1D clue: Less colorful
Answer: PALER

2D clue: Layer of the upper atmosphere
Answer: OZONE

3D clue: Totally pointless
Answer: NOUSE

4D clue: Hit a high note in a high place, perhaps
Answer: YODEL

5D clue: Kit ___ bar
Answer: KAT

How to play more Mini Crosswords

The New York Times Games section offers a large number of online games, but only some of them are free for all to play. You can play the current day’s Mini Crossword for free, but you’ll need a subscription to the Times Games section to play older puzzles from the archives.

Technologies

Want to Speak to Dolphins? Researchers Won $100,000 AI Prize Studying Their Whistling

The scientists studied a bottlenose dolphin community in Sarasota, Florida, uncovering evidence of language-like communications.

If any dolphins are reading this: hello!

A team of scientists studying a community of Florida dolphins has been awarded the first $100,000 Coller Dolittle Challenge prize, set up to award research in interspecies communication algorithms.

The US-based team, led by Laela Sayigh of the Woods Hole Oceanographic Institution, found that a type of whistle that dolphins employ is used as an alarm. Another whistle they studied is used by dolphins to respond to unexpected or unfamiliar situations. The team used non-invasive hydrophones to perform the research, which provides evidence that dolphins may be using whistles like words, shared with multiple members of their communities.

Capturing the sounds is just the beginning. Researchers will use AI to continue deciphering the whistles to try to find more patterns.

«The main thing stopping us cracking the code of animal communication is a lack of data. Think of the 1 trillion words needed to train a large language model like ChatGPT. We don’t have anything like this for other animals,» said Jonathan Birch, a professor at the London School of Economics and Politics and one of the judges for the prize.

«That’s why we need programs like the Sarasota Dolphin Research Program, which has built up an extraordinary library of dolphin whistles over 40 years. The cumulative result of all that work is that Laela Sayigh and her team can now use deep learning to analyse the whistles and perhaps, one day, crack the code,» he said.

The award was part of a ceremony honoring the work of four teams from across the world. In addition to the dolphin project, researchers studied ways in which nightingales, marmoset monkeys and cuttlefish communicate.

The challenge is a collaboration between the Jeremy Coller Foundation and Tel Aviv University. Submissions for next year open up in August.

Dolphins are just the beginning

Researching animals and trying to learn the secrets of their communication is nothing new; but AI is speeding up the creation of larger and lager datasets.

«Breakthroughs are inevitable,» says Kate Zacarian, CEO and co-founder of Earth Species Project, a California-based nonprofit that also works in breaking down language barriers with the animal world.

«Just as AI has revolutionized the fields of medicine and material science, we see a similar opportunity to bring those advances to the study of animal communication and empower researchers in this space with entirely new capabilities,» Zacarian said.

Zacarian applauded Sayigh’s team and their win and said it will help bring broader recognition to the study of non-human animal communication. It could also bring more attention to ways that AI can change the nature of this type of research.
«The AI systems aren’t just faster — they allow for entirely new types of inquiry,» she said. «We’re moving from decoding isolated signals to exploring communication as a rich, dynamic, and structure phenomenon — whish is a task that’s simply too big for our human brains, but possible for large-scale AI models.»

Earth Species recently released an open-source large audio language model for analyzing animal sounds called NatureLM-audio. The organization is currently working with biologists and ethologists to study species including carrion crows, orcas, jumping spiders and others and plans to release some of their findings later this year, Zacarian said.