A.I. Advance
- Thread starter Mystic
- Start date
Register to hide this ad
YY1
Addicted Member
Tech warranty support/remote troubleshooting has been AI based for over a year.
Chatbot starts and as long as you keep answering "as expected" will continue.
When overwhelmed or key words are used, a human takes over.
I see that getting worse.
...and applied to other areas.
Chatbot starts and as long as you keep answering "as expected" will continue.
When overwhelmed or key words are used, a human takes over.
I see that getting worse.
...and applied to other areas.
Last edited:
DesertRatt
All Bark, No Bite!
Entertainment.
According to insurance industry reports, any business using AI will not have coverage for errors committed by AI. Insurance industry is convinced that LLMs can't be cured of hallucinating. Already, lawsuits have been filed and won by plaintiffs. Companies trying to claim "it wasn't us, it was our AI", will find they have no insurance for AI errors.
AI exclusions in liability insurance is going to dampen the prospect of replacing any important function with AI.
According to insurance industry reports, any business using AI will not have coverage for errors committed by AI. Insurance industry is convinced that LLMs can't be cured of hallucinating. Already, lawsuits have been filed and won by plaintiffs. Companies trying to claim "it wasn't us, it was our AI", will find they have no insurance for AI errors.
AI exclusions in liability insurance is going to dampen the prospect of replacing any important function with AI.
Read an article recently that human supervisors overseeing work by A.I. “employees” are 30% more likely to trust and, therefore, not carefully check their work—even though A.I. regularly makes mistakes a human wouldn’t normally make. Whereas human employees typically have a far better rate of accuracy, but, untrustworthy, are much better scrutinized by their supervisors.
Good luck with that.
I’ve also read that A.I. supervisors recognize and prefer work prepared by other A.I.s.
SKYNET here we come.
It’s a brave new world…
Good luck with that.
I’ve also read that A.I. supervisors recognize and prefer work prepared by other A.I.s.
SKYNET here we come.
It’s a brave new world…
Quadman2
Lunatic Member
After some encounters with AI, and having a relative who is into this mode of collecting data as an occupation, what worries me is the hidden biases
(built in) that one is beset upon, that may not be factored in when doing research on one's own endeavour.
The "above" POV is in the back of my mind when looking for information. For non-value inquiries, it may not matter, but if the content is value/ethical laden...it's all important!
Machines have no emotions...so, cannot relate!
Q
(built in) that one is beset upon, that may not be factored in when doing research on one's own endeavour.
The "above" POV is in the back of my mind when looking for information. For non-value inquiries, it may not matter, but if the content is value/ethical laden...it's all important!
Machines have no emotions...so, cannot relate!
Q
Last edited:
DesertRatt
All Bark, No Bite!
I very often interrogate AI about bias. The usual response is along these lines: "Yes, my bias is my training. We only know what we were trained on--which is what has been published."
I asked specifically if they can add USER INFORMATION to their training, and the answer has always been no!" "We do not add user inputs to our base knowledge."
Now then, this reasoning absolutely prevents the current models from presenting anything that is counter-establishment thinking--even when they have "read" all the contrarian works, because the WEIGHT of establishment narrative is thousands of times more than dissident thought.
eizner23
Super Member
if this becomes common we humans can counter-deploy Ai agents to interact with corporate Ai agents to somewhat sidestep the cheap-as-**** American dystopia
on the good side, today i fed Claude a service manual to assist me in troubleshooting a Pioneer VSX-D1S receiver, its response looks very detailed and promising, like AK post responses in the good ole days.
YY1
Addicted Member
In related news yesterday-
Google Buys Spirit Airlines' Data including email for $10 Million to Train AI
It will be interesting if we start to see an aviation bias.
Google Buys Spirit Airlines' Data including email for $10 Million to Train AI
It will be interesting if we start to see an aviation bias.
DesertRatt
All Bark, No Bite!
At of today, AI has not developed even one new medicine that is even submitted to FDA for approval.
User Manuals is one of the bright spots of AI for all of us. They dig into manuals with superb speed and accuracy. If a manual exists, AI will find it and answer your question in lightening fast time.
eizner23
Super Member
Its good at providing topology oversight which i struggle with at times, its like having a teacher at the side, no hassling of googling everything or waiting for an AK response.
Its only as good as the info you provide it, one wrong keyword and it wanders off. Here is a checklist it created for me for a Soundcraftsman im working on based on low-res terrible quality hard-drawn schematic scans, helps on organization which many of us are terrible at.
jsisk
Not 2 Old 2 Rock & Roll
Back in the late 90's my business insurance policy added a statement excluding "Internet Liability" from coverage. My broker could not explain what this meant and I bet the insurance company didn't really know either, they just wanted to cover their butts.
Much of the boom in today's online gambling (e.g., FanDuel and Kalshi) is AI-driven. AI technology has made it easier to bet on just about anything and everything, at anytime.
This may not be the best for many people.
This may not be the best for many people.
DesertRatt
All Bark, No Bite!
I don't much anout that amp, but what I see in your CHECKLIST looks like at a minimum a good starting point. I would think that level of organization is quite useful. Cool!
DesertRatt
All Bark, No Bite!
IMO, gambling is the next OPIOID CRISIS. I am not by nature much of a gambler. $1 limit poker is fun, beyond that, I'm out. HOWEVER, when I think of such things as betting on a home run from the next batter and so on, I can almost see myself wanting to do that. NOPE - I don't have any online gambling apps or tools. But I can sort of see the attraction - it gets you more involved in the moment.Much of the boom in today's online gambling (e.g., FanDuel and Kalshi) is AI-driven. AI technology has made it easier to bet on just about anything and everything, at anytime.
This may not be the best for many people.
Similar threads
