1650429101344169984

@MikhailBurtsev @yurakuratov Shame we don’t understand well enough how human long-term memory works. It would be interesting to try and calculate the total number of tokens that a human can “store” in a lifetime. Or, perhaps, the number of tokens that constitute his/her personality and core beliefs, etc.

1650430078558281730

See? This is why we need #StableLM and other open source AI models. So we can transparently test the implications of this before a mass deployment inside Office or whatever. https://t.co/LNjKWS9lpw

1650431630786396160

The smartest artists are the ones that use AI to multiply their reach, not the ones that sue to death. https://t.co/oCUEi8rArI

1650430638355300352

Now the question is: can you build a complete personality with 2M tokens?

If yes:
Job of the future: synthetic personality designer.

If no:
How many more tokens are necessary? https://t.co/LNjKWS9lpw

1650419957312700422

Something to think about deeply as you start the week:

Some GPT-4 users will soon get a version of the AI with a much larger “memory”. It might cost a bit more, but it’ll be able to carry on a coherent conversation for hours or it’ll be able to write multi-page documents in a… https://t.co/U1iO0fwaGW

1650405384224374786

This is what I mean when I say that, in AI, almost every week, you go to bed and, the morning after, you discover that somebody has announced some major progress in this or that area and you have to read and test a ton of stuff.

What’s the takeaway? Don’t go to bed. https://t.co/6BLYjUU8DY

1650406024753426433

This is what I mean when I say that, in AI, almost every week, you go to bed and, the morning after, you discover that somebody has announced some major progress in this or that area and you have to read and test a ton of stuff.

What’s the takeaway? Don’t go to bed. https://t.co/LNjKWS9lpw

1650260992297783299

Step 1: stop talking to people next to you IRL to talk to people somewhere else on the planet.
Step 2: stop talking to people next to you IRL *and* to people somewhere else on the planet to talk to a chatbot pretending to be other people.
Step 3: stop talking?

1650195331101913090

Weeks ago, Seth Godin wrote:

“If your work isn’t more useful or insightful or urgent than GPT can create in 12 seconds, don’t interrupt people with it.”

https://t.co/Hx28VOCFBb https://t.co/v59DABcVEx