Thank you, Erik and Gerben, for a beautifully explained and elaborated piece. For a (passably!) numerate layman like me, it was engaging and very interesting. It also provided a lot of detail and context, which helped to “flesh out” the intuitive misgivings I’ve had, arising from the disconnect between hyperbole and actual “product”.
I’ll be watching with interest as we continue, as a society, to obsess over our “flawed mirror”.
You're welcome, Richard. And really getting reader response like this is just awesome. Gerben took care to put that together, and his YouTube talk is really worth watching. Many thanks.
The talk itself takes more time to consume and people (myself included) are loathe to blindly invest that much time. But it is a more 'complete & coherent' package and I hope people actively help distributing it. I would live for it to be so widespread that it is able to push back a bit against the onslaught of poor quality information.
Many thanks, Gerben : I anticipate that I’ll watch the whole YouTube talk, as, to me, it’s an important topic and being better informed about limitations helps to cut through some of the current hyperbole surrounding LLM AIs such as ChatGPT.
I look forward to seeing it and really appreciate the time and effort that folks such as yourself and Erik are putting into clarifying things.
Thanks so much for this. I have been engrossed as of late in the discussion of how AI will change healthcare, and the conviction, to use your apt word, reminds me of speaking to a group who recently converted to a new religion—true believers, to borrow Hoffer's parlance.
This video, along with Erik’s work, is so important in building a cogent counternarrative built on more than cynicism. I especially appreciated the quotes below, which acknowledge LLM’s benefits but also warn that decisions have consequences, as inertia and pain of “sunk costs” can prevent future change.
“…My advice would be: to concentrate on establishing realism about the use cases….”
But,
“If you wed yourself to these kinds of huge systems, you've actually really fixed yourself to something which is very hard to change.”
Thank you, Erik and Gerben, for a beautifully explained and elaborated piece. For a (passably!) numerate layman like me, it was engaging and very interesting. It also provided a lot of detail and context, which helped to “flesh out” the intuitive misgivings I’ve had, arising from the disconnect between hyperbole and actual “product”.
I’ll be watching with interest as we continue, as a society, to obsess over our “flawed mirror”.
You're welcome, Richard. And really getting reader response like this is just awesome. Gerben took care to put that together, and his YouTube talk is really worth watching. Many thanks.
The talk itself takes more time to consume and people (myself included) are loathe to blindly invest that much time. But it is a more 'complete & coherent' package and I hope people actively help distributing it. I would live for it to be so widespread that it is able to push back a bit against the onslaught of poor quality information.
Many thanks, Gerben : I anticipate that I’ll watch the whole YouTube talk, as, to me, it’s an important topic and being better informed about limitations helps to cut through some of the current hyperbole surrounding LLM AIs such as ChatGPT.
I look forward to seeing it and really appreciate the time and effort that folks such as yourself and Erik are putting into clarifying things.
Thank you. If you like what you read/see, please actively spread it as it is very hard to get heard in a sea of hype
Thanks so much for this. I have been engrossed as of late in the discussion of how AI will change healthcare, and the conviction, to use your apt word, reminds me of speaking to a group who recently converted to a new religion—true believers, to borrow Hoffer's parlance.
This video, along with Erik’s work, is so important in building a cogent counternarrative built on more than cynicism. I especially appreciated the quotes below, which acknowledge LLM’s benefits but also warn that decisions have consequences, as inertia and pain of “sunk costs” can prevent future change.
“…My advice would be: to concentrate on establishing realism about the use cases….”
But,
“If you wed yourself to these kinds of huge systems, you've actually really fixed yourself to something which is very hard to change.”
Thanks again for the post.
Alan
This is really encouraging Alan. Thank you so much.
Beautifully done. I will distribute to students and others. Thank you.
Wonderful, David. Thank you.