The popular LLMs are generalist showpieces that stumble over domain-specific tasks.
(In many cases a homegrown ML model and/or code would do a better + faster + cheaper job.)
I think he's already home ...
(btw for those who haven't heard of drug smugglers' techniques, the short version is:
these groups will out-innovate a corporate team any day of the week.
If it moves across borders, they have someone figuring out how to get drugs into it.)
I sometimes joke that drug smugglers' techniques will soon be used to move other goods.
But I didn't think "Nvidia chips" would be on that list.
The problem, and a possible solution, by @mileskellerman.bsky.social :
Gains from genAI adoption are both overstated and unevenly distributed.
(Even worse: Of the relatively few companies that could benefit from the technology, many will fail to do so because they lack the requisite expertise, time, money, and discipline.)
Q McCallum
Q McCallum
Q McCallum
Q McCallum
Q McCallum
Today's episode of "(Perhaps) Technically Correct But Still The Wrong Thing To Say Out Loud"
