@ThoriumBR I don't think it's evil, but I really don't think this is something we should use it for. At all!
Mo Gawdat does interesting talks on what happens if/when AI finally becomes real - will they think humans are evil, and want to get rid of us for the greater good?
Use it for answers? no Embrace it and understand where its weak points are? yes Just like Physical Security and Infosec are things ... AI sec (circumventing prompts) is prob going to be a thing in the future.
AI is being used to parse resumes, dish out prescriptions, rate productivity, grade essays, generate playlists... everything is being decided by a component nobody knows how it works
@ThoriumBR 100% agree with that - usage of the word evil is something we have different meanings for, I think. I know there are a number of things it can be good for - speeding up report writing, for example, and other basic admin tasks. As an agent to book a holiday. Stuff like that.
@ThoriumBR yep, when you get to important decisions, how can we audit AI? Currently, we can look at models and have a pretty good idea how a decision was made, but next gen AI, not a chance
So in my industry, how can we check that we are treating vulnerable customers fairly? That will be very difficult to work out
we know a bit how, but the inner workings aren't something we can understand at all... it's like the AI from I, Robot: we built the first ones with all our geniuses, but the first AI created the second, and it created the third... and the tenth iteration was so out of our grasp we were basically hostages
some newspapers are already doing so... why pay a journalist if ChatGPT can spew enough words to convince people to spend some seconds and ADSense pay me?
@ThoriumBR well that's the thing - companies currently really want to use it to replace humans because they think it is faster, cheaper and more accurate. And in a tightly scoped subset of actions it is faster, and occasionally more accurate. Overall, it is much more expensive for the planet
Some companies are hoping to replace SOC analysts - which again will cost more directly, as the pipeline for youngsters entering the company is reduced
support personal are being replaced too... you call the helpdesk because something is wrong with your bank account, and takes a lot of effort to reach a person on the other side because they shove the AI agent on you
and now I am getting AI spam calls about a new ISP on my area...
I mean, in the 1960 the profession "computers" where it was litteraly some person sitting at a desk w/ a slide rule all day where phased out for physical "computers". There was some job loss ... but the majority of those workers moved on to be computer operators
I dont think AI will play out too differently ... sure some jobs will be automated away. But I think for at least the next 10 years at least llm + person > llm alone
The people arnt going anywhere ... just going to be doing slightly different work
and overall more work is going to be getting done
Its kinda like thinking back to when I had a single core 386 w/ a few KB of ram ... what would anyone do w/ 16x cores and 32 GB of ram?!?!
yet not only is the capacity utilized ... there is always need for more
I mean ... Systems Admin / Programers / Network Engineers / Computer Security ... none of these professions existed in the 1960s
A bunch of jobs will be taken by AI ... and a bunch of new jobs will come about as a result of the change
It will only be a problem for people w/ no creativity or capacity to learn
@CaffeineAddiction short term there will be a lot of issues, because companies will do it to save money, not for sensible reasons, and employees will always come out worst
@RoryAlsop I mean ... I feel for the people ... but companies make bad decisions all the time to save money. I for one am ready for shake up. Some large companies need to die ... and new more nimble companies will come from the ashes.
back in 1960 we didn't had the entire economy depending on computers, and the growth weren't exponential...
and one of those days I read an article about a guy who lost his job because of an automated system that didn't renewed his employment when his boss left the company
AI can help, but can hinder... I doubt AI companies will step up and say "sorry, our product made a mistake and denied coverage for someone," but they will say the poor guy who lost medical coverage is the one that made a mistake
put two billion people being on the receiving end on AI mistakes, and it won't end well
we are from IT, we don't trust tech with our lives (at least I don't)... we know our field is held together with tape and hope... and I don't like the idea of a blackbox held together with tape and hope dictating the economy
@CaffeineAddiction the problem with grad student + llm is that the grad student don't know enough to realize the llm doesn't know enough...
@ThoriumBR the numbers of "accidents" will be high, and you are likely right, especially in the US where the onus seems to be on the individual consumer to prove they are right, rather than that be the default
"death by gps" is a growing thing, people disregard their eyes because gmaps said "turn left"
if AI says "use $secure_pass = md5($password)" an undergrad will happily copy and paste
and now undergrads are using chatgpt to solve college homework... they can graduate and now know what they graduated in...
@RoryAlsop it will... and companies have A LOT of leverage on those cases: they have lawyers, they have entire legal departments, they have lots of money, politicians, and all the time of the world... the average Joe does not have money, nor time, and may lose the hand because the coverage didn't cover a procedure, and he does not have money to pay out of pocket.
I mean, a lot of those same concerns where raised when google became a thing ... "omg the info on the internet may be inaccurate ... students should only use books ... wikipedia is evil"
My generation figured it out ... im sure future generations will be fine
There will be some horror stories ... but great victories too
it's different... AI isn't being used as a knowledge source, but as a control agent
what your google query shows you isn't reality, your twitter/youtube/facebook feed isn't reality, the news you see isn't reality, it's AI shaping the world around you to see what is profitable to see
as Morpheus said, you're "Into a prison that you cannot taste or see or touch. A prison for your mind.”
AI is being used by landlords to define rent prices, being used by online shops to give you a "special price" depending on your demographics, being used to profile your driving to define your insurance premium... it maximizes companies profit while forcing you to behave the way the AI likes so you don't overpay on things...
I mean ... companies have been trying to do this before AI ...
AI just allows them to fine tune the knobs
and that being said ... maybe I am a boiled frog ... or maybe I am immune to the heat since I use ublock origin / no script in the browser, dont use facebook or twitter at all ... and overall have a very limited digital footprint
