Correct me if I’m wrong but my understanding is most everything ChatGPT can do was already possible with GPT3 (especially post InstructGPT) but it just took more intentional wrangling. What ChatGPT seems to be offering is a much more accessible interface.

Thanks!

Thanks so much for doing this!

Nitpick: the "advice I've given people so far" link is broken.

We didn't run a draft of this post by DM or Anthropic (or OpenAI), so this information may be mistaken or out-of-date. My hope is that we’re completely wrong!

Why not run a draft of the post by them? Not sure what you had to lose there and seems like it could've been better (both from a politeness/cooperativeness perspective and from a tactical perspective) to have done so.

I don't feel personally affected by this change but this seems to matter a lot to other people, for example:

• If there is any chance you'll publish your work in an academic journal, or otherwise will be publishing it somewhere else in the future... not all publishers will care about this license, but it might be prudent to be cautious because some definitely will.

• If you do not want podcasts or translations of your work to be made without your consent (e.g. maybe you'd want to work with any translators to ensure accuracy — you now won't be able to control this).

• If you do not want your work potentially to be used for commercial purposes.

• If your work is done for contract, and not funded by a grant, you'll need to review the contract terms and depending on the terms of the contract, you may need to get consent from the funder prior to posting.

• The CC BY 4.0 license is irrevocable, so even if you think there is a chance of your mind changing with regards to one of the above, you could be screwed.

I think people posting deserve to make an informed choice. I'm happy to see the new posting includes a checkbox "Before you can publish this post you must agree to the terms of use including your content being available under a CC-BY license" - this satisfies me. But still possible people might not fully understand the implications so I encourage people to be thoughtful about this.

Came here to say this! EA Twitter is fun!

I don't think these Twitter polls are at all helpful and risk being misleading.

This is not just because they are highly non-representative as you admit but also there is no pre-event test to compare results to so it's hard to draw conclusions about how the FTX situation contributes to the distribution of results you did get. It's also impossible to know how the first poll connects to the second poll as they could be going to somewhat different audiences (for example - the first question got more votes) and you can't do crosstabs between the two questions.

Rethink Priorities is doing some work looking into this and we could very easily and quickly do more work with representative samples if further commissioned to do so. It's a bummer we don't have EA Pulse running yet as this would be an ideal use case.

Thank you for sharing. I found this helpful.

For me personally, it's hard for my primary emotion to not be anger - anger at Sam Bankman-Fried and his inner circle for recklessly gambling away the future of the world.

not fine IMO

This is so ironic.