I Think AGI Just Happened

https://danielmiessler.com/blog/i-think-agi-just-happened

by dimava • 5 days ago

4 4 5 days ago

Yeah...AGI isn't going to successfully replace Chris. But you know what this shit will do? It'll teach people that any information that they share about what they're doing or how they do it will ultimately be weaponized against them.

That'll work out well in the long term.

jzu48• 5 days ago

Its not about what its doing well. Its about what and where it screws up. The older you get you stop extrapolating from what you see any tool doing in the demo or marketing ppt or the first month. Just as with people. This is why we have probation periods for employee. It takes Time to see how they handle all the unpredictability. There will be screw ups. And then you write your blog post.

More complex a system gets, more issues it has. Never less.

gnabgib• 5 days ago

And how to I make blueberry pie?

anenefan• 5 days ago

Well then, let us pretend whatever might qualify as AGI is every bit as capable of as a regular employee that's going to get the job. A job which first requires screening to get that job and an expectation any applicant should be able to work just as well or better after successfully getting the job. AI might be suitable for analysis, ... various but limited , though possibly very technical and involved job functions -- AGI however would need to be as good as any generally capable person given time to perform the task ... any ... not limited to something a smart determined team could code a script to do ... or merely replacing cookie cuter type jobs - however AI is probably a better choice for some of the cheap cookie cutter tech support that's been offered over the years.

Consider seriously, how many AIs could match a generalist's ability and knowledge and would have a job after a few months such as if they got the job of research assistant. I can say I would have fired those AIs I've encountered thus far that work in the public arena - either as use for populating wikis or a blog or offering search answers - too many pitiful errors ... a couple of honest ones is forgivable, a handful is a problem that needs sorting out, correct only 95% of the time, bye.

What for example some laypersons seem to be learning lately by way of web, is a good indication of poor AI adoption. AI is wonderful for patterns, statistics and endless fruitful possibilities. But with information a single mistake might be overlooked, it happens -- what doesn't pass is that type of bloke we've all met somewhere ... sometime in our lives, who seems to know everything, a wealth of information, right up until they stumble onto something we ourselves are specialists in, and we come to slowly realise they are no more than a skilful BS artist - even not half of what they are talking about turns out to be BS ... 10% or 5% BS is still infuriating as it takes more effort to correct everyone else who's been hanging on their words.

I Think AGI Just Happened

Related Stories