[ExI] all we are is just llms was: RE: e: GPT-4 on its inability to solve the symbol grounding problem

[spike at rainier66.com]

…> On Behalf Of William Flynn Wallace via extropy-chat
Subject: Re: [ExI] all we are is just llms was: RE: e: GPT-4 on its inability to solve the symbol grounding problem

 

 

>>…we don’t know how to measure ourselves, even by our own standards.

spike

 

>…No, you are not getting away with that one no matter how great the song is (acoustic guitars played excellently).

 

Cool, did I accidentally write a theme for a country song?  I honestly do not know to what song you refer, Billw.  Please offer a link.

 

>…  If they didn't, industrial psychologists would never ever be hired again because their predictions don't result in better hiring, for one example.    billw

OK cool, I propose a contest of sorts.  Let’s have ChatGPT interact with a transcript of the applicants for a position, have it predict the success of the applicants.  Do not let the hiring managers have access to GPT’s opinion.  At the same time, have another manager at the same company read the transcript and make her own assessment, but let both the other manager and GPT have the same info: text only.  After a year or two, have the manager rank the new-hires, compare GPT’s prediction on the suitability of the applicant.

It is easy to imagine ChatGPT eventually figuring out how to be a better hiring manager than its carbon-based counterpart.

spike

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20230420/27e23add/attachment.htm>