@caseynewton@zoeschiffer It's 2024, and Google's Big Plan for the Future of the Internet is to spend hundreds of billions of dollars training an algorithm to recreate AOL.
So interesting to read the dystopic conclusion to Ben Thompson's framework of "Platform vs Aggregators" in the age of Gen AI: for the aggregator to squeeze the market so hard that it hurts itself.
Looks like we're walking towards the absolute opposite of the Web 3.0 dream (not the crypto bullshit, the semantic web)... all knowledge gobbled up in an inscrutable database, with zero discoverability for the little gems out there.
@caseynewton great coverage of OpenAI! Regarding employee equity, how is OpenAI offering stock options? I thought it is a non-profit that has a close relationship/JV with an LLC controlled by Microsoft.
@caseynewton OpenAI will eventually release a bio and pictures of the unnamed voice actress, and entirely coincidentally, she will look startlingly similar to Scarlett Johansson.
@caseynewton I wouldn't buy that "moving with caution" propaganda either, much like their pretense at being "open" and a non-profit was merely a way to avoid regulation and to pretend their aims weren't nefarious.
@caseynewton Probably typical old man syndrome on my part but the number of companies that seems to start with "don't be evil" and then are shocked when VC money or going public forces "profits at any cost" has become a pretty exhausting merry-go-round
@caseynewton@zoeschiffer My physics professor taught us to watch for words like "obviously follows" and "of course." He said this is how a lecturer elides the part of the proof they don't understand.
Up to this point, I was wondering why you didn't find anyone from Facebook to talk to—even though FB, much more than Amazon, Apple, and Google, was roiled by employee dissent in 2020, and before that in 2018 in the wake of the Cambridge Analytica scandal and the Kavanaugh hearings. Now I see why.
@caseynewton "...a company that once proudly served as an entry point to a web that it nourished with traffic and advertising revenue has begun to abstract that all away into an input for its large language models."
Curious to see if a LLM, once built, can sustain itself without new human generated content. Why build your new 'fields of dreams' on the web, if users are never going to come anyway.
@caseynewton@mike They’re taking the web out behind their server farm shed to shoot it.
Meanwhile none of the other AI stuff (“organize your receipts”) actually works well enough to be useful because you have to second-guess it constantly; this has been done before and more accurately with boring algorithms like OCR and string matching.
@caseynewton Hmmm, not sure that is a good idea, you could already tweak the responses a bit. Out of curiosity I asked ChatGPT to call me dude, and insert scifi and heavy metal references. Which made the answers completely useless. It got so tangled up in its responses that facts mattered less than those quotes.
Fun but useless. Add some force to your code and rock on. Surf’s up dude!