I kind of tried to replicate your experiment (in German where "Erdbeere" has 4 E) that went the same way. The interesting thing was that after I pointed out the error I couldn't get it to doubt the result again. It stuck to the correct answer that seemed kind of "reinforced".
It was also interesting to observe how GPT (4o) even tried to prove/illustrate the result typographically by placing the same word four times and putting the respective letter in bold font (without being prompted to do that).
I haven't played with this model, but rarely do I find working w/ Claude or GPT-4 for that to be the case. If you say it's incorrect, it will give you another answer instead of insisting on correctness.
If you find a case where forceful pushback is sticky, it's either because the primary answer is overwhelmingly present in the training set compared to the next best option or because there are conversations in the training that followed similar stickiness, esp. if the structure of the pushback itself is similar to what is found in those conversations.
I'll put it another way - behavior like this is extremely rare in my experience. I'm just trying to explain if one encounters it why it's likely happening.
> OpenAI will receive access to current and archived content from News Corp’s major news and information publications, including The Wall Street Journal, Barron’s, MarketWatch, Investor’s Business Daily, FN, and New York Post; The Times, The Sunday Times and The Sun; The Australian, news.com.au, The Daily Telegraph, The Courier Mail, The Advertiser, and Herald Sun; and others. The partnership does not include access to content from any of News Corp’s other businesses.
These publications aren't exactly known for their level-headed, unbiased journalism. What a shame that we're poisoning the well and allowing old media to maintain a stranglehold on crafting their own narrative.
It’s naive to take the charter of openai at face value. It’s hilarious to see applications of non-profit tax filings as I’ve grown up. There is tremendous wealth and commerce flowing through the US tax free. Religions, universities, “good” tech companies. To whose benefit? Humanity’s? Lol
I bet the nonprofit with a small for profit arm that the tech is licensed from is the next best thing since the defeated incorporating in Ireland for tax avoidance.
Title is misleading. This article is about the NHTSA investigating the efficacy of Tesla's recall last year. There is no mention of "hundreds of crashes"
Then surely all the good actors have to do KYC, and all the bad actors can just pretend to be American entities.
I don't agree with this on principle, but even just from a practical perspective it seems like they are leaving the door completely open by doing that. What's even the point?
