Impromptu - Reid Hoffman - Review


Reid’s Conflicts of Interest

First off we need to state the conflicts of interest:

Reid is on the board of MSFT, had early access to GPT4 and is an investor in several firms that are building or building on LLMs. That said Reid, who is a partner at Greylock Ventures, has stepped down from the non-profit board of OpenAI to avoid conflicts of interest recently.

Understandably Reid espouses what I deem to be overly techno-optimistic views. GPT4 of course helps but his questions are, quite often, leading in nature i.e “How will X benefit from LLMs” vs “How will X be impacted by LLMs”

Start by reading this

  1. GPT4 writes a plot twist - Pg51-55 - stellar example of creativity on tap - simply blew me away
  2. The conundrum with apprenticeship jobs - Pg 128-130 - illustrated with the example of management consulting: if management consulting firms aren’t hiring armies of bright young university and B-school graduates to perform the basic tasks of management consulting, how will the next generation of partners learn their craft?

This is true of lots of other professions that groom from within and where knowledge grunt work is treated as a rite of passage - I-Banking, Investing, Software Development and much more

Some useful insights / arguments:

  • Having GPT4 complete a joke that Reid had earlier administered to GPT3 and seeing the difference in quality was his Newton-apple moment. (I cannot agree more)
  • Wikipedia is imperfect and yet useful. Why not LLMs? (Fair point)
  • AI content creators will earn millions. So what? There are cats that have youtube channels with a million subscribers today. (Touche)

Strong Chapters:

  • Moments of Enlightenment
  • Hallucinations
  • Useful Operating Principles

Weak Chapters: (Worrying because the arguments for techno-optimism are quite weak imo)

  • Journalism
  • Transformation of Work

Here is the link to the book (pdf version)

Go check it out!