Fwd: More AI links for taco list from Hal

[Nan and Sandy Sanders]

>
>Forbes has a few good short podcastsstle articles  worth the ride...
>"Chat GPT4 Is 5X Smarter Than chat GPT3: Tech Icons Launch Petition to Pause"
>"GPT-4: Commotion and Controversy"
>"GTP-3 Release: What it means For the Future of Your Business"
>(You may hear the name Hinton, he is considered the father of the 
>deep learning).
>
>Letter from AI luminaries Asking for a Pause.... (avoid the press 
>hype, this is the actual letter)
>https://futureoflife.org/open-letter/pause-giant-ai-experiments/
>
>One thread of the debate.  (Scott is a well known quantum complexity 
>theorist currently on leave working at OpenAI on "alignment" -- the 
>field trying to ensure
>that a sufficiently powerful AI doesn't decide to eliminate humanity)
>https://scottaaronson.blog/
>Follow his back and forth with comment authors too.
>Don't feel bad if you can't follow the quantum complexity 
>discussions.   I advise staying clear of those.
>
>Hallucination.   A problem with all AI based on deep learning and 
>traced to "semantic noise" in the training data and the deep 
>learning model itself.
>
>"ChatGPT Invents a sexual harassment scandal and names a real law 
>professor as the accused"
>https://www.washingtonpost.com/technology/2023/04/05/chatgpt-lies/
>
>Hallucination  is a serious problem.  Hallucination shows up 
>everywhere in deep learning AI.  It has been traced to what is 
>called "semantic noise" in large training data.
>ChatGPT is text, but there are others that identify things in a 
>image (self-driving car) , perform medical diagnosis, decide if a 
>jailed individual is dangerous ( the so called "psychotic test") and 
>should be denied parole.
>
>How about using it to predict propagation?
>
>
>
>
>
>
>