The newest AI wonder, ChatGPT — the latest in a line of incredibly quickly-evolving AI text generators — is causing jaws to drop and brows to furrow.
What's happening: Users are telling ChatGPT to rewrite literary classics in new styles or to produce performance reviews of their colleagues, and the results can be scarily good.
Why it matters: ChatGPT displays AI's power and fun. It could also make life difficult for everyone — as teachers and bosses try to figure out who did the work and all of society struggles even harder to discern truth from fiction.
Driving the news: Last week's public release of ChatGPT came from OpenAI, which had previously set benchmarks in this field with GPT3 and its predecessors. (There's also an unofficial Twitter bot for those who don't want to bother with signing up for the service.)
- Early tech adopters went wild requesting and sharing stories, jokes and poetry, such as this bible song about ducks and an amazing sonnet on string cheese.
Yes, but: The high quality of ChatGPT's responses adds to the fun, but also highlights the risks associated with AI.
- As we just wrote last week, a big pitfall for today's most advanced AI programs is their ability to be "confidently wrong," presenting falsehoods authoritatively.
- That's certainly the case with ChatGPT, which can weave a convincing tale about a completely fictitious Ohio-Indiana war.
- Nightmare scenarios involve fears that text from AI engines could be used to inundate the public with authoritative-sounding information to support conspiracy theories and propaganda.
- OpenAI chief Sam Altman says some of what people interpret as "censorship" — when ChatGPT says it won't tackle a user request — is actually an effort to keep the bot from spewing out false info as fact.
Between the lines: ChatGPT, like other text generators, also creates problems when it gets things right. Educators, who already often have to run essays through online tools to make sure they weren't plagiarized, worry that their difficult task could be made even harder.
- Altman said on Twitter that OpenAI is working to address a wide range of feedback. One interesting question that he notes has arisen is whether powerful AI like ChatGPT should evolve to behave the way users want or stay true to the way its creators intend.
Zoom in: I gave ChatGPT a few tasks on Sunday, with varying success.
- First I asked it to write an article on ChatGPT in Axios style, because that would have saved me a ton of time. It did fine summarizing its own capabilities, but knows nothing about Axios' style.
- Then I asked it to write a performance review for my excellent editor. It said it didn't know him well enough to do a personal review, but did offer some generic positive comments that I probably could have used without anyone knowing the difference.
- I asked it to write a rap about me and I have to say, it's more flattering than some of the pictures of me I created with the viral AI app Lensa.
Zoom out: Even in its present form, ChatGPT can serve up useful answers to plenty of questions — and that's without being trained on the latest news and information.
- With some refinement, it could easily become a great search interface, as some have pointed out. (Google, along with others, is already aiming to answer more search queries directly.)
What they're saying: Though people are clearly fascinated by ChatGPT, opinion is decidedly mixed on its net impact.
- Box CEO Aaron Levie: "ChatGPT is one of those rare moments in technology where you see a glimmer of how everything is going to be different going forward."
- PyTorch co-creator Soumith Chintala: "ChatGPT seems to be **really** good for creative work and a solid starting point for mundane work (similar to CoPilot). It is unlikely i will trust it with automation, where you need predictability. I wish in the next iterations, they hook it up to verification systems."
- Ars Technica AI reporter Benj Edwards: "For now, it’s possible that OpenAI invented history’s most convincing, knowledgable, and dangerous liar — a superhuman fiction machine that could be used to influence masses or alter history. But I applaud their cautious roll-out. I think they are aware of these issues."
