Where’s Waldo? The Elephant in the Room
We recorded 9 claims from this post. 3 were checked against evidence. 6 can’t be checked. 3 more are held. Read the post at its source.
Claims from this post (9)
In the order they appear in the post. Each quotation is the speaker’s words in this post. A label such as “Supported” or “Can’t be checked” is the claim’s current assessment; its page has the evidence and the reasons.
Assessment: Can’t be checkedSaid
“Here is a lovely and instructive set of examples just posted on X by the data scientist Colin Fraser, which serve to remind us that the beauty of LLM’s remains skin deep:”
What we tested: The apparent quality of LLMs is only skin deep.
Assessment: UnresolvedSaid
“Examples like these, in which ChatGPT confidently generates an image that does not actually meet requirements, are nicely complement the visual hallucinations that Ernest Davis and I reported here in October, in which visual ChatGPT misinterpreted various images:”
What we tested: In examples like those shown, ChatGPT confidently generated images that did not actually meet requirements, and in examples reported in October, visual ChatGPT misinterpreted various images.
Assessment: Can’t be checkedSaid
“In some ways, the most telling example in the earlier newsletter was from Dave Andress (which Vicki Bier shared with us), highlighting the fact that image generators still don’t really understand how the world works:”
What we tested: Image generators still do not really understand how the world works.
Assessment: Can’t be checkedSaid
“When you have terabytes of images in your training set you can generate really cool images, but that doesn’t guarantee that your system will understand the relation between those images and the world.”
What we tested: Training on terabytes of images enables impressive image generation but does not guarantee the system understands how images relate to the world.
Assessment: UnresolvedSaid
“Such errors clearly aren’t disappearing any time soon.”
What we tested: Errors of discomprehension in neural-network-based systems will not disappear any time soon.
Assessment: Can’t be checkedSaid
“Sometimes GenAI used for source attribution will work, other times it will probably yield outrageous goofs, same as we have seen over and over in other domains.”
What we tested: Generative AI used for source attribution will sometimes work but will probably sometimes produce outrageous errors.
Assessment: SupportedSaid
“We can of course expect related problems of discomprehension in efforts to use these systems to address copyright issues, as I wrote earlier this week.”
What we tested: Efforts to use generative AI systems to address copyright issues will run into problems of discomprehension.
What happened: The public report documents a completed evaluation using generative vision-language models for copyright detection. The event date is the release of that empirical report; individual testing dates are not specified. (arXiv; Xu and colleagues, )
Assessment: Can’t be checkedSaid
“Unfortunately, errors of discomprehension may soon be even more worrisome in a new context: war.”
What we tested: AI errors of discomprehension may soon become more worrisome as AI is used in war.
Assessment: Can’t be checkedSaid
“Not sure that’s a great idea, in an era in which AI still can’t quite sort out what’s happening with an elephant on a crowded beach.”
What we tested: Marcus is not sure that hooking AI up to military applications such as drones is a great idea in an era in which AI still can't quite sort out what's happening with an elephant on a crowded beach.
Held from this post (3)
These records did not pass every check, so nothing about them is published. Held is not a finding.
- The criteria are ambiguous. Our reading, not assessed: OpenAI appears to be opening its systems to military applications, and at least one well-funded startup is connecting AI to military drones.
- The criteria are ambiguous. Our reading, not assessed: Statistical approximation through generative AI is rarely reliable.
- The criteria are ambiguous. Our reading, not assessed: AI still can't quite sort out what's happening with an elephant on a crowded beach.