Hacker News Reader

Comment by consf | Hacker News Reader

I think a good marketing move is showcasing the seamlessness of artificial intelligence. For some reason, AI is being stigmatized

View on HN

Yeah, a really basic and relevant example might be if I have a meeting coming up on my calendar, an agent can pull info from my LinkedIn about the participants, check their socials (so I can make small talk and chat), research info about some of the topics in the meeting (e.g. pull stats, recent articles, etc.) and summarize it to prepare me for the meeting.

Like, that's actually useful but probably not super sexy.

This is a good example of where we should have stigma attached to AI: a partner showed me their CRM that summarized my profile for them. It was creepy in how far into my online life it had dug. It also said, along with the creepy things, that I was "risk adverse" and "only acted when I had complete information" ... as a startup CTO this is both laughably wrong and could have career implications.

LLM's can be tuned to do what you want it to do. It can be as simple as "summarize these notes" to as questionable as "rate this candidate for this job".

The former is mostly just a reduction in task and condensation of information. The latter is the opposite: an extrapolation of some conclusion based on given information.

This is not the fault of an LLM, but the people using it and the use cases that have been designed. An AI assistant that just gave me the facts (condense) rather than generate its own conclusions (expand, extrapolate) is fine, IMO.

I get what you are saying but, the same way ad targeting can be used to "serve me more interesting ads" and "surveillance" can be used to find missing children and sex trafficing victims, it isn't being used that way across the board.

Reply on news.ycombinator.com