What a grand act of service you have performed here! The response of ChatGPT-4 shows that programming is still at the level of conscience. Such legislation as the program itself recommends is an imperative.
Bill Gates has been involved in AI for years. Hardly the paragon of perfection, Gates's Windows programs were typically released FAR too early. I know because I was a volunteer beta-tester for Win 97, Win NT (later Win 2000), and all of them in the early stages of development were horrid, and were still horrid after release. (The first version of the beta Win97 restarted every single time a driver loaded, and was still iffy in the 3rd and 4th iterations.) His name on AI gives me significant pause; I still don't trust his ability to opt for self-control over "just get it out there." [I realize Microsoft is no longer controlled by Gates, FWIW, but his interest in AI goes waaayyy back, and included sneaking into AI conferences in disguise.]
Part of the problem with IT is the enthusiasm for the "newest, latest, greatest" but little or no ethical considerations. Microsoft's products, at that time, were so bad I finally started using Linux and/or FreeBSD on my computers just to be able to work and kept the MS product on a partitioned hard-drive.
My suggestion is that the Gates Foundation ought to found and fund an international, NEUTRAL, ethics review board, get nations to sign on and adhere to it, before unleashing this any farther to the public. The Gates Foundation wants to be seen as an innovative do-good body, HERE is a good place to show that.
Thank you for elevating this discussion. I really don't think there is much serious discussion occurring amongst the congresscritters, especially since so many of them are deeply techno-ignorant (remember the "inter-tubes"?).
We KNOW that the tech industry cannot be trusted to look out for the future. Their focus on the future gets REALLY fuzzy past the next fiscal quarter or year. However, they will be shoveling millions of $$ of bribes ("campaign contributions") to the congresscritters to make sure that any meaningful regulation is tied up indefinitely. We still don't have anything close to reasonable privacy protection in the U.S. for this very reason.
Speaking of lack of privacy protection, the potential to misuse of AI to fraud & identity theft is massive. With no accountability or transparency, it's going to get pretty scary pretty soon.
Finally, a topic that rarely gets mentioned, but the writers on the picket lines have brought it up a few times - AI could accurately be called a "plagiarism BOT" in that it takes creations that are on the Internet (which is nearly everything) and typically without attribution or credit "creates" something new. Again, no accountability or transparency. . .
We have some dark times ahead . . I agree that we're not approaching a Skynet inflection point yet, but primarily due to hardware limitations. Let's hope no one thinks that putting this into a orbital defense platform is a good idea . . .
For those of us of the generation where we hand-wrote letters, used rotary phones and watched the first color tv, this article is incredibly helpful! And deeply unsettling. Thank you!
Recently I retired. I was a History major. Do I need to go back to school to comprehend AI and CHAT GPT. I do not think my brain can absorb this concept. It’s frightening. I had my annual mammogram recently. My report was great, no signs of breast cancer. However the scan was read “by a computer”. This bothered me. Dr. Eric Topol at the Scripps Institute (His Substack is “Ground Truths”) feels very positive about CHAT GPT. I feel threatened.
I’m like huh!? Maybe we could ask Ai to rewrite the tax laws for a more equitable and prosperous economy, or for legislation to provide improved healthcare outcomes for everyone or to protect our environment?
On another forum, I asked whether, if there had been AI as we are envisioning it to be in Einstein's day, whether it would have come up with the Special Relativity or would it have been constrained by all the stuff it could download about Newtonian physics? The general thought was that there is still something uniquely human about a leap of the imagination.
I know Chatbot can write a good essay, and I'll bet it can write good standard formulaic fiction the way that thousands of fleshly AIs on Amazon Unlimited can already do. But can it write a good non-formulaic novel? Or non-derivative poetry? Could it have dreamed up The Wasteland? Can it figure out how to look at character outside of stereotyped ways? Can it empathize? How do you program for compassion?
Is a leap of the imagination required for sentience? Can AI become our Overlord even if it can't think its way out a huge box created by all the ideas thought up till now? Part of the problem is that I don't think we really know how imagination operates in the human mind. "Bright ideas" seem to just happen.
Your comments regarding ChatGPT are cogent and astute, however, I must disagree with your basic assumption that Artificial Intelligence exists and is demonstrated by ChatGPT. ChatGPT is an app written by programmers, and as such all of its functions are known and predictable in the grand traditions of Wordstar, Lotus 1-2-3, or DB2.
More to the point. engineering works when all facts are known and the parameters can be quantifiably defined. It's the practical application of science.
Science can be seen as research, but it's a process for fact-finding. You gather known facts, and form a hypothesis, an untested theory based on your facts. A lot of people think the process stops there, but you then test your hypothesis to see if it works. Whether it works or not, the information generated by your testing is therefore used to generate new tests and new hypotheses. And so it goes.
If your hypothesis is rigorously tested and found to be sound under a variety of conditions, then it may eventually become a theory. An extremely-well tested theory is called a scientific law. Newton's Laws of Motion have undergone centuries of testing and research, and are still found to accurately describe bodies at rest and in motion. By contrast, Einstein's famous equations are still regarded as a Theory.
Science has failed the search for Artificial Intelligence because it has not provided engineers with the design and definition of intelligence. This means engineers have nothing to design. As far as I can tell from reading articles in popular media, the field is balkanized into a million research labs protected by nondisclosure agreements and guarded by lawyers. If there's a working theory to the field, they're all trying to build bigger databases, because eventually someone's gonna build a big enough database that a miracle will happen and the computer will become intelligent.
My final point here is that not even the firm making ChatGPT is claiming their product is intelligent. ChatGPT is a big database of information, and it's algorithm is able to make creative connections with the words it has. But if this were actually science, someone would be able to look at a display and recite a list of working functions the system is accomplishing, and categorically state that the definitions of intelligence have been accomplished. As it is, nobody seems to know why it works.
We're in the position of owning a dictionary and hoping that if you add enough words and pages to it, the book will suddenly leap to sentience.
I'm not a computer scientist or even related to one. I've read a lot about A{, and think my talking points come from a laypersons nontechnical point of view.
I have been an avid sci-fi fan since the mid-60’s. I threw myself into the open thinking that Star Trek provided. I have said all along that if sci-fi writers can think of it, ‘it’ can become real. All of it, the good and the bad.
This very sudden jump into public AI is no different and, viewed with current social issues with the truth, we must get ahead of it with firm boundaries that even the [corrupt] SCOTUS cannot undo.
This was a wonderfully useful piece. It occurs to me to ask AI for possible solutions to the issues central to The Big Picture. Like, when does a human fetus become a person?
Hmmm. Invent machines that can essentially think for themselves, give them no guardrails and access to them by some truly bad people.
Why, what could possibly go wrong????????
What a grand act of service you have performed here! The response of ChatGPT-4 shows that programming is still at the level of conscience. Such legislation as the program itself recommends is an imperative.
Bill Gates has been involved in AI for years. Hardly the paragon of perfection, Gates's Windows programs were typically released FAR too early. I know because I was a volunteer beta-tester for Win 97, Win NT (later Win 2000), and all of them in the early stages of development were horrid, and were still horrid after release. (The first version of the beta Win97 restarted every single time a driver loaded, and was still iffy in the 3rd and 4th iterations.) His name on AI gives me significant pause; I still don't trust his ability to opt for self-control over "just get it out there." [I realize Microsoft is no longer controlled by Gates, FWIW, but his interest in AI goes waaayyy back, and included sneaking into AI conferences in disguise.]
Part of the problem with IT is the enthusiasm for the "newest, latest, greatest" but little or no ethical considerations. Microsoft's products, at that time, were so bad I finally started using Linux and/or FreeBSD on my computers just to be able to work and kept the MS product on a partitioned hard-drive.
My suggestion is that the Gates Foundation ought to found and fund an international, NEUTRAL, ethics review board, get nations to sign on and adhere to it, before unleashing this any farther to the public. The Gates Foundation wants to be seen as an innovative do-good body, HERE is a good place to show that.
This needs to he published far and wide!
Fascinating read.
Thank you for elevating this discussion. I really don't think there is much serious discussion occurring amongst the congresscritters, especially since so many of them are deeply techno-ignorant (remember the "inter-tubes"?).
We KNOW that the tech industry cannot be trusted to look out for the future. Their focus on the future gets REALLY fuzzy past the next fiscal quarter or year. However, they will be shoveling millions of $$ of bribes ("campaign contributions") to the congresscritters to make sure that any meaningful regulation is tied up indefinitely. We still don't have anything close to reasonable privacy protection in the U.S. for this very reason.
Speaking of lack of privacy protection, the potential to misuse of AI to fraud & identity theft is massive. With no accountability or transparency, it's going to get pretty scary pretty soon.
Finally, a topic that rarely gets mentioned, but the writers on the picket lines have brought it up a few times - AI could accurately be called a "plagiarism BOT" in that it takes creations that are on the Internet (which is nearly everything) and typically without attribution or credit "creates" something new. Again, no accountability or transparency. . .
We have some dark times ahead . . I agree that we're not approaching a Skynet inflection point yet, but primarily due to hardware limitations. Let's hope no one thinks that putting this into a orbital defense platform is a good idea . . .
For those of us of the generation where we hand-wrote letters, used rotary phones and watched the first color tv, this article is incredibly helpful! And deeply unsettling. Thank you!
Recently I retired. I was a History major. Do I need to go back to school to comprehend AI and CHAT GPT. I do not think my brain can absorb this concept. It’s frightening. I had my annual mammogram recently. My report was great, no signs of breast cancer. However the scan was read “by a computer”. This bothered me. Dr. Eric Topol at the Scripps Institute (His Substack is “Ground Truths”) feels very positive about CHAT GPT. I feel threatened.
I’m like huh!? Maybe we could ask Ai to rewrite the tax laws for a more equitable and prosperous economy, or for legislation to provide improved healthcare outcomes for everyone or to protect our environment?
On another forum, I asked whether, if there had been AI as we are envisioning it to be in Einstein's day, whether it would have come up with the Special Relativity or would it have been constrained by all the stuff it could download about Newtonian physics? The general thought was that there is still something uniquely human about a leap of the imagination.
I know Chatbot can write a good essay, and I'll bet it can write good standard formulaic fiction the way that thousands of fleshly AIs on Amazon Unlimited can already do. But can it write a good non-formulaic novel? Or non-derivative poetry? Could it have dreamed up The Wasteland? Can it figure out how to look at character outside of stereotyped ways? Can it empathize? How do you program for compassion?
Is a leap of the imagination required for sentience? Can AI become our Overlord even if it can't think its way out a huge box created by all the ideas thought up till now? Part of the problem is that I don't think we really know how imagination operates in the human mind. "Bright ideas" seem to just happen.
This, to me, is more frightening than a DeSantis presidency!
Sirs:
Your comments regarding ChatGPT are cogent and astute, however, I must disagree with your basic assumption that Artificial Intelligence exists and is demonstrated by ChatGPT. ChatGPT is an app written by programmers, and as such all of its functions are known and predictable in the grand traditions of Wordstar, Lotus 1-2-3, or DB2.
More to the point. engineering works when all facts are known and the parameters can be quantifiably defined. It's the practical application of science.
Science can be seen as research, but it's a process for fact-finding. You gather known facts, and form a hypothesis, an untested theory based on your facts. A lot of people think the process stops there, but you then test your hypothesis to see if it works. Whether it works or not, the information generated by your testing is therefore used to generate new tests and new hypotheses. And so it goes.
If your hypothesis is rigorously tested and found to be sound under a variety of conditions, then it may eventually become a theory. An extremely-well tested theory is called a scientific law. Newton's Laws of Motion have undergone centuries of testing and research, and are still found to accurately describe bodies at rest and in motion. By contrast, Einstein's famous equations are still regarded as a Theory.
Science has failed the search for Artificial Intelligence because it has not provided engineers with the design and definition of intelligence. This means engineers have nothing to design. As far as I can tell from reading articles in popular media, the field is balkanized into a million research labs protected by nondisclosure agreements and guarded by lawyers. If there's a working theory to the field, they're all trying to build bigger databases, because eventually someone's gonna build a big enough database that a miracle will happen and the computer will become intelligent.
My final point here is that not even the firm making ChatGPT is claiming their product is intelligent. ChatGPT is a big database of information, and it's algorithm is able to make creative connections with the words it has. But if this were actually science, someone would be able to look at a display and recite a list of working functions the system is accomplishing, and categorically state that the definitions of intelligence have been accomplished. As it is, nobody seems to know why it works.
We're in the position of owning a dictionary and hoping that if you add enough words and pages to it, the book will suddenly leap to sentience.
I'm not a computer scientist or even related to one. I've read a lot about A{, and think my talking points come from a laypersons nontechnical point of view.
I have been an avid sci-fi fan since the mid-60’s. I threw myself into the open thinking that Star Trek provided. I have said all along that if sci-fi writers can think of it, ‘it’ can become real. All of it, the good and the bad.
This very sudden jump into public AI is no different and, viewed with current social issues with the truth, we must get ahead of it with firm boundaries that even the [corrupt] SCOTUS cannot undo.
This was a wonderfully useful piece. It occurs to me to ask AI for possible solutions to the issues central to The Big Picture. Like, when does a human fetus become a person?