+- Carlton Supporters Club (http://new.carltonsc.com)
+-- Forum: Social Club (http://new.carltonsc.com/forum-6.html)
+--- Forum: Blah-Blah Bar (http://new.carltonsc.com/forum-23.html)
+--- Thread: AI and creativity. (/thread-6302.html)
Re: AI and creativity. - Lods - 10-12-2023
Don't the 'Doomsday' predictions rely largely on a lack of vigilance on the part of humans at every stage of the AI progress and development.
If the buggers start their shenanigans we'll just take out their batteries
Re: AI and creativity. - kruddler - 10-12-2023
(10-12-2023, 01:58 AM)Lods link Wrote:Don't the 'Doomsday' predictions rely largely on a lack of vigilance on the part of humans at every stage of the AI progress and development.
If the buggers start their shenanigans we'll just take out their batteries
Initially i did say without the lack of proper fail-safes. But it only takes one mistake, or one revenge driven former employee, or one Dr Evil to circumvent that.
I suspect that these will be running on a technology that we are not using yet, or at least is just theoretical.
Remember this stuff probably wont happen in our lifetimes. Batteries may be as 'old hat' then as horse driven carts are now. Around for novelty purposes only.
Re: AI and creativity. - Thryleon - 10-12-2023
(10-12-2023, 01:58 AM)Lods link Wrote:Don't the 'Doomsday' predictions rely largely on a lack of vigilance on the part of humans at every stage of the AI progress and development.
If the buggers start their shenanigans we'll just take out their batteries
IF, THEN, AND, OR, ELSE statements.
That's all computers really do. IF variable condition with value open exists, THEN perform X AND/OR Y AND/OR Z, activities until variable condition changes to value close. Or something like that.
Do they learn? Well, you could argue that it will create a record of each outcome it performs an action for, and then review how to stop something from happening but it will be performed with another IF statement. Data corruption is more likely to end this whole thing than any other factor. Unplug the data and you unplug the ability for it to be intelligent.
What will quantum computing do? The above sort of parameters is capable of running in microseconds as it is, but the slow bit is the pipeline in, the collection of outcomes, and then after a certain number of outcomes have been collected, looking at them to see which ones are repeated, and what else could be done to stop the outcome from occurring. We are talking a few seconds to minutes of actions being collected, analysed and repeated all with scripting and coding.
The idea that they learn is not the same as someone learning something. We learn that fire is hot, and it is to be feared because when you get that feeling of going near something hot, it could hurt you. That instinct is not what the computer learns. It will learn a sequence of actions that get taken when something is hot to stop the thing from getting hot, or it will implement a sequence of actions to avoid the hot. It doesnt learn this, it finds that information and parrots it. You wipe the disk? The memory is gone? You have the smartest brick on the planet.
Where it gets murky is if someone starts leveraging AI without telling people and safe guards are built in to prevent things from happening, but ive seen far too many systems run into trouble because of stupid stuff.
i.e. A business builds a quantum computer server farm. This leverages all the hardware into a common purpose. That farm is unregulated but able to start doing whatever with only a few people's input. Form then on, how this pans out is largely dependant on whom is in control of that technology. Isolating it from the web is easy, but if it can resolve those issues and bypass security measures itself, then it opens itself up to all manner of input. Once it has the data how it will act is a bit of an unknown known. The programs and conditions it has committed to memory it will enact until conditions change, but the scripts it runs will largely be predetermined.
What Kruddler states about the technology is a mixed bag. I don't think we are going to see 2001: A space odyssey type stuff. The computers arent capable of rational, or irrational thought. They just perform instructions.
Case in point. At my place of employment, we run virtual servers in a microsoft data centre. They had an issue where one of the hosts had an issue and tried to fail us over. The operating system failed to start, and I had to restore the virtual machine from a backup and then perform remedial action to get its SQL server database back to normal. Luckily it was part of a pair, and the primary was happy as larry, and this was the secondary. It wasnt smart enough to realise that the cluster was repaired and to start reconciling the changes made on the primary to the secondary again without human interaction, and it was incapable of restoring itself from backup without interaction from us.
THAT is where these things are not that smart. According to Microsoft, the thing was fine and healthy but in reality the VM was in a failed to start mode that appeared fine.
Re: AI and creativity. - kruddler - 10-12-2023
Thry,
What you are talking about is basic computer programming. Done by humans, with fail safes involved and best of intentions.
New 'deep learning' or 'machine learning' doesn't work the same way.
These 'teach themselves' and we don't understand how they do it.
There are some basic examples of this, and some very advanced stuff.
For instance, there was one that was developed to 'model and predict' how some type of pendulum or rotationary machine worked.
As humans, we can describe it, model it and explain it.
The computer worked out its own way of modelling it and we don't know how it did it, but we know its different to our way.
It gets the right answer, we just have no idea how.
...and it worked it out by itself.
....and thats a very basic example.
I previously mentioned there is a type that can diagnose diseases from x-rays far better than humans. Not only that, it can do it on x-rays that have the resolution of something like PONG. As humans, we could barely work out what part of the body the x-ray is of, AI not only knows, but can predict what is cancerous with greater accuracy than the best trained doctors with the best technology.
Again, we don't know how.
These are very basic, just starting out levels of AI that are already showing up humans. How advanced will it be 20 years from now?40 years from now? 100 years from now?
20-25 years ago, i was trying to download an mp3 file and my 4hour dial-up reset would click in before that 6mb file would download....and i'd have to try again.
Now i can watch, in HD, a live video of same song/concert in real time, while other people in the same house are doing the same with their own interests being fulfilled equally well.
People are traditionally bad at predicting the future. Partially because the jobs of the future don't actually exist currently.
We project our own knowledge and expectations into the future, without reflecting on what occured over that same time period in the past.....and with the knowledge of technology increasing exponentially in the meantime.
Re: AI and creativity. - Thryleon - 10-12-2023
(10-12-2023, 05:03 AM)kruddler link Wrote:Thry,kruddler I'm talking about how this works.
What you are talking about is basic computer programming. Done by humans, with fail safes involved and best of intentions.
New 'deep learning' or 'machine learning' doesn't work the same way.
These 'teach themselves' and we don't understand how they do it.
There are some basic examples of this, and some very advanced stuff.
For instance, there was one that was developed to 'model and predict' how some type of pendulum or rotationary machine worked.
As humans, we can describe it, model it and explain it.
The computer worked out its own way of modelling it and we don't know how it did it, but we know its different to our way.
It gets the right answer, we just have no idea how.
...and it worked it out by itself.
....and thats a very basic example.
I previously mentioned there is a type that can diagnose diseases from x-rays far better than humans. Not only that, it can do it on x-rays that have the resolution of something like PONG. As humans, we could barely work out what part of the body the x-ray is of, AI not only knows, but can predict what is cancerous with greater accuracy than the best trained doctors with the best technology.
Again, we don't know how.
These are very basic, just starting out levels of AI that are already showing up humans. How advanced will it be 20 years from now?40 years from now? 100 years from now?
20-25 years ago, i was trying to download an mp3 file and my 4hour dial-up reset would click in before that 6mb file would download....and i'd have to try again.
Now i can watch, in HD, a live video of same song/concert in real time, while other people in the same house are doing the same with their own interests being fulfilled equally well.
People are traditionally bad at predicting the future. Partially because the jobs of the future don't actually exist currently.
We project our own knowledge and expectations into the future, without reflecting on what occured over that same time period in the past.....and with the knowledge of technology increasing exponentially in the meantime.
https://www.sas.com/en_au/insights/analytics/what-is-artificial-intelligence.html#:~:text=AI%20adapts%20through%20progressive%20learning,product%20to%20recommend%20next%20online.
Theres nothing that intelligent about it. It's all about the information you feed it.
Re: AI and creativity. - kruddler - 10-12-2023
(10-12-2023, 09:04 AM)Thryleon link Wrote:kruddler I'm talking about how this works.
https://www.sas.com/en_au/insights/analytics/what-is-artificial-intelligence.html#:~:text=AI%20adapts%20through%20progressive%20learning,product%20to%20recommend%20next%20online.
Theres nothing that intelligent about it. It's all about the information you feed it.
Read the last part of that....
Quote:In summary, the goal of AI is to provide software that can reason on input and explain on output. AI will provide human-like interactions with software and offer decision support for specific tasks, but it’s not a replacement for humans – and won’t be anytime soon.
Based on CURRENT forms of AI.
Think longer term.
Here is a few quotes i found from people you may recognise...
“The pace of progress in artificial intelligence (I’m not referring to narrow AI) is incredibly fast. Unless you have direct exposure to groups like Deepmind, you have no idea how fast—it is growing at a pace close to exponential. The risk of something seriously dangerous happening is in the five-year time frame. 10 years at most.”
—Elon Musk wrote in a comment on Edge.org
“The development of full artificial intelligence could spell the end of the human race….It would take off on its own, and re-design itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded.”
— Stephen Hawking told the BBC
“I don’t want to really scare you, but it was alarming how many people I talked to who are highly placed people in AI who have retreats that are sort of ‘bug out’ houses, to which they could flee if it all hits the fan.”
—James Barrat, author of Our Final Invention: Artificial Intelligence and the End of the Human Era, told the Washington Post
“I’m increasingly inclined to think that there should be some regulatory oversight, maybe at the national and international level, just to make sure that we don’t do something very foolish. I mean with artificial intelligence we’re summoning the demon.”
—Elon Musk warned at MIT’s AeroAstro Centennial Symposium
There are a lot of people smarter than you or I than can see this as becoming an issue and i don't see any reason to doubt them.
Re: AI and creativity. - LP - 10-12-2023
Yes, neural networks and quantum computing rely on carefully crafted questions to deliver accurate responses.
Google conned the public a few years back claiming it's quantum engine solved a question in seconds that would take a conventional supercomputer decades. But the bogus nature of the claim was hidden in the detail. The Google quantum computer was designed and optimised to answer that specific question, in fact that is all it could do, and they used the unoptimised worst case scenario model for the Supercomputer.
If similarly optimised the Supercomputer would take about 2hrs, but the supercomputer can also answer General Purpose computing questions, while the quantum computer could not.
At the moment and into the foreseeable future quantum computers are task specific, like Hino selling one truck to carry bricks, and another truck that carries timber.
Re: AI and creativity. - Thryleon - 10-12-2023
(10-12-2023, 09:29 AM)kruddler link Wrote:Read the last part of that....
Based on CURRENT forms of AI.
Think longer term.
Here is a few quotes i found from people you may recognise...
“The pace of progress in artificial intelligence (I’m not referring to narrow AI) is incredibly fast. Unless you have direct exposure to groups like Deepmind, you have no idea how fast—it is growing at a pace close to exponential. The risk of something seriously dangerous happening is in the five-year time frame. 10 years at most.”
—Elon Musk wrote in a comment on Edge.org
“The development of full artificial intelligence could spell the end of the human race….It would take off on its own, and re-design itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn’t compete, and would be superseded.”
— Stephen Hawking told the BBC
“I don’t want to really scare you, but it was alarming how many people I talked to who are highly placed people in AI who have retreats that are sort of ‘bug out’ houses, to which they could flee if it all hits the fan.”
—James Barrat, author of Our Final Invention: Artificial Intelligence and the End of the Human Era, told the Washington Post
“I’m increasingly inclined to think that there should be some regulatory oversight, maybe at the national and international level, just to make sure that we don’t do something very foolish. I mean with artificial intelligence we’re summoning the demon.”
—Elon Musk warned at MIT’s AeroAstro Centennial Symposium
There are a lot of people smarter than you or I than can see this as becoming an issue and i don't see any reason to doubt them.
I'm still talking about how this works on a functional level.
The hello world component of it all. Function, algorithm, patterns, data, logic.
Apply that to people. They don't respond anywhere near as consistently because emotions, attachments, void of logical, empathy.
The hard disk is still just a library of instructions, which vary based on data, logic, and outcomes.
I've heard what you've stated. I see what those above have written and predicted. To a degree, uncontrolled I think it could get out of hand, but it would have to be so far removed from our current concept of a computer and how they work that it still isn't AI but a computer drawing bad conclusions, from faulty modelling. The exact opposite of intuitive knowledge and ergo not intelligent.
Re: AI and creativity. - LP - 10-12-2023
AI as it is currently being touted is dependant on LLM, massively parallel datasets that are trolled by massively parallel neural networks in the form of a Generative Adversarial Network, to construct algorithms and relationships between bits of data. But the LLM can't distinguish between reality and fantasy, and the dataset is not trivial. For example you could craft a question that would goad an AI into answering a question with facts relating to how bones of fire breathing dragons were found amongst dinosaur fossils.
Quantum computers are at the moment limited to a few hundred qubits, and they give instant answers to carefully crafted "questions", (This question is typically a conjecture expressed in math) but that questions can take weeks, months or years to get correct, and then take weeks, months or years to form a proof that both the question and the answer were correct. Think of the movie "The Man Who Knew Infinity."
At the moment Nano whatever works down to a molecular level, they are singular function devices that have zero intelligence or awareness.
That last word awareness is the critical term, because without awareness any system will be easily deceived no matter how well we craft it to be perceived as intelligent, and that is why our lauded AI LLM wonders can inform you "There be Dragons!"
If you want to worry about a technology going forward, probably nefarious enhanced humans might be your biggest threat, a uber brain in a complete bastard!
Re: AI and creativity. - LP - 10-16-2023
This AI debate reminded me of a lecture from Hannah Fry a few years back, I think part of the Royal Society Christmas Lecture series.
This video and others well worth watching;
https://www.youtube.com/watch?v=Rzhpf1Ai7Z4