FabSwingers.com > Forums > The Lounge > Thoughts on AI
Thoughts on AI
Jump to: Newest in thread
| |||
| |||
| |||
| |||
"I believe that it is an overblown thing.People trying to find too much significance in technology that while impressive in some has very extreme limitations.Limitations that take time to sort But I have been wrong before.Very wrong Your thoughts? " I recently wrote a paper on OSINT for cyber defense. It touched on AI several times. If you think about what AI is, it's not intelligent at all, and it's not artificial, it's real. My understanding of the name is that it came from some sort of futuristic concept from the 70's. If you look at the likes of Google and Bing, what do their AI systems do? Google and bing; their job is to get some text from you, and offer you lists of websites that seem to be related. The more one thing gets clicked on, the higher it ranks, because the person who typed the term found that particular result as helpful, etc. so Google deems that as more likely to be "the answer". It's basically a list of indexed web pages. Thele role of AI in these systems is to add another layer on top of that. When you type a search into AI, an automation system does the same as before, but this time, it does many searches, it finds related buzz words from a list. It searches through pages on your behalf, extracts bits according to various algorithms, and then presents them to you along with the related links. If you think about it, there's nothing intelligent about that, and it's not artificial. It's just the next layer of work that can be afforded to use now that libraries are more developed, and not that processing power and storage aren't really hard to come by. I wrote an AI system about 10 years ago. It had a light sensor, an RTC, relays and lots of other things. It could make decisions based on various sensors. It was good, but it wasn't intelligent and it wasn't artificial. I understand that nowadays I there are some things that cause alarm. Apparently Israel has some AI guided bombing system. No doubt Bae are probably involved in some way. The UK is at the cutting edge of this intelligence gathering thing. They set up the first OSINT system just after WW1, so that it could lead the way for automation. | |||
"I think the AI we have at the moment is very simplistic. That’s not to say it will stay that way of course. As the uses and intelligence become greater, one stumbling block is morals. Whose morals do “robots” get taught? For example, a self driving car is driving down the street when a 50yo man steps out in front of the vehicle. It can either hit the man, swerve left and hit a child or swerve right into oncoming traffic with an unknown outcome. Someone has to preprogram a “moral” code for the vehicle to base its decisions on. The famous Isaac Asimov I Robot with its 3 basic laws of robotics are a good start but it will be so much more complex that that at some point. " It's not a moral issue when someone gets mowed down. No one made a choice to mow them down and a self driving car would probably be better equipped than the average human at detecting and avoiding hazards. | |||
| |||
| |||
| |||
| |||
| |||
| |||
| |||
| |||
"I think the AI we have at the moment is very simplistic. That’s not to say it will stay that way of course. As the uses and intelligence become greater, one stumbling block is morals. Whose morals do “robots” get taught? For example, a self driving car is driving down the street when a 50yo man steps out in front of the vehicle. It can either hit the man, swerve left and hit a child or swerve right into oncoming traffic with an unknown outcome. Someone has to preprogram a “moral” code for the vehicle to base its decisions on. The famous Isaac Asimov I Robot with its 3 basic laws of robotics are a good start but it will be so much more complex that that at some point. It's not a moral issue when someone gets mowed down. No one made a choice to mow them down and a self driving car would probably be better equipped than the average human at detecting and avoiding hazards. " The choice of which action to take is moral. In the absence of instructions as to what to do, a computer does nothing. So the computer driving the car has to be given instructions as to what to do in 1000s of different circumstances. That’s basically what a computer program is, a list of “if” questions and an answer. If a man steps into the road, brake No time to brake, swerve If there is kid in the way, brake No time to brake, swerve Sometimes there isn’t an outcome that isn’t bad. Most humans would instinctively avoid the man and the kid and swerve into oncoming traffic. But that’s the option with the highest risk of personal harm. You don’t have time to think of the consequences, you just react. A computer can’t do that, it has to weight up the options based on the information it has. It has to be told which is the “best” outcome, and that’s a moral decision. | |||
| |||
| |||
"With the possibility of AGI in the next two years the implications for rapid change are staggering. News about artists and coders losing jobs is only scratching the surface. Academia and research in every scientific field is going to be completely reshaped and accelerated by orders of magnitude. Weapons of the kind we haven't imagined yet are no doubt already being planned and we will have new kind of nuclear nonproliferation treaty's. Autonomous robots will eventually become the norm in service, labour and military applications. Once AGI gives way to ASI all bets are off. No one can predict what will happen after that. Jeffery Hinton, who was awarded the Nobel prize this year for developing technology that made AI possible is worth reading. Ray Kurzwiel is worth a look too. Those are two of the guys that have brought AI to where it is today." I know they are also working on a super computer atm too. What takes people minutes, or ai seconds, will allow ai do compute complex things in milliseconds. It will allow for advanced AI learning and the such. I like the ai of it used for NPC ai in games. Like I seen it used as a mod for Skyrim, where NPCs could reference the games lore to answer the player. Slow and clunky, but cool. | |||
| |||
| |||
| |||
"Artificial intelligence will never be a match for natural stupidity." That comment wins the Internet for today  | |||
| |||
"A dancing bear. The implications of the current gen AI / LLM’s are the levelling up of the unskilled and semi Skilled that work in tech and admin , it’s already happening, the democratisation of bottom end skills in writing. Real AI hasn’t been invented yet. The current systems take the output of human learning. There are no systems yet synthesise the actual learning and reasoning process. " I wouldn't say that's entirely correct. Systems exist today that have surpassed humans on the Abstract Reasoning Benchmark. | |||
| |||
