Musk And The AI Letter

I agree with what you say. But when I read the warnings of Musk/Google/Apple/Facebook I mostly read: "Wah, we were extracting rent from our AI advancements and now we got leapfrogged by other companies that we don't own or extract rent from! Wah! Legislators make them stop! Think of the shareholders! We can't be wasting money on innovation!"

Maarten Daalder

Nothing I hate more than a customer service BOT!

Siliconself

What happens when so few can afford the products that the bots are making that profits start to fall.

Greg Richard

Some are only saying the brakes should be applied, are doing so while they catch up.

Simon York

Michael F Barushok... Thanks for your questions. Please let me break it down a bit: 1. We all should be making ethical decisions - personal and vocational ones, regardless of who stands to benefit from them. Ethics shouldn't be compromised. 2. Those that wish to harm others with their poor ethics, will do so with or without AI, so the argument that AI would lead to ethics violations follows in my next point. 3. As with most "advancements", or changes in opportunities made available by Tech and even Nature, said changes are like double edged swords. I would hate to think that we are always willing to toss out the baby with the bath water, because we are not prepared to deal with opportunities in some form of Best Practice bylaws and legislative regulations. AI in and of itself need not be a problem if the coding used follows rules that preclude AI from causing injury. Just like it's pretty difficult to divide by zero, or use a month date beyond standard calendar date rules. 4. In fact, I think in many instances, AI can do a superior job over humans because of their objectivity. AI, given at least the rule sets that humans use to make decisions, should be able to make some pretty good decisions without being influenced by greed, or prejudice. A system for "Degree of Confidence" can be established just as it is used in statistical analysis. If that degree falls within a parameter that needs further review, then so be it, and assign that task to those who are best able to make that call.... a single person, or group decision that is sized appropriately. 5. What guidelines do we use when our decisions are based on empathy. Do we ever act against our feelings for empathy? I think humans ignore empathy a great deal of the time. We still have wars. We still have murders. We still make the lives of many miserable, even when we know better because of that tug on oiur conscience. That tug on our empathy. But I doubt AI would do likewise in those cases. How would AI deal with Mercy Killing of a terminal patient? Could AI be programmed not to toss a delinquent paying tenant out of a shelter for failure to pay rent due to circumstances? I think it could be taught to weigh the facts for an appropriate suggestion to the property owner. True, ethics can be a sticky widget, but I think guidelines can be codified in much the same way that we use guidelines when we decide upon ethical choices. As with anything that poses potential danger of any sort, oversight and regulations will need to be in place to mitigate any injuries from the potential danger, but I doubt that would mean disbanding AI all together. Just have controls put into place ... Ask AI to help formulate those oversight, regulatory controls designed to mitigate injury when AI is in service. Is empathy a learned behavior in the same way that most human behavior is? If so, then teach it to AI.

Ethical decisions cannot be objectively entrusted to those who benefit from the decision. Ethical dilemmas are even more difficult, as those who aren't directly benefiting might be motivated by minimizing their losses to the detriment of many others. Philosophical reasoning aside, would humans ever trust that unemotional decisions are superior to decision making requiring empathy?

Michael F Barushok

Fair points, relative to today's innovations, but your second point touches on where I was going with my thoughts when I asked, when would AI be found to be in violation of Asimov's Rules for Robotics? I think it can be proposed that at some point, the use of AI could have an injurious effect on humans - especially in the jobs field, which leads to more pain... so at that point, AI would be harming humans... maybe not by their own animosity toward humans, but what if AI were put in charge of assigning tasks, and it determined that no human participation was necessary, so it allocated all tasks to AI? Is that an action that needs to be ruled out from happening, because the resulting layoffs would likely harm the human employees and probably their families as well. Harm can come in many forms. The Asimov's Rules for Robotics restrict harm that might be inflicted on humans by robots. The key is in the definition of "Harm by Robots", and can AI be defined as robotic - since robots perform as programmed, and so does AI.

If regulation is needed, it's not on the job replacement. It's when other organisations try to save money by automating official decisions, court decisions etc. That's where the danger lies : not in the their strengths, but in their weaknesses.

adrian

But don't over-react to what these things are doing. They can parse natural-language queries and compose replies in a convincing way. That's it. There's no thought or reasoning involved, just analysis according to programmed factors and predictive-text responses. If added to a puppet like many current robots, they can replace a TV presenter or a telephone answering machine. If they make decisions, they are merely amplifying the biases of their creators.

adrian

Of course, that doesn't mean we shouldn't think about it, whether we want to industrialise forms of work that we haven't done previously, how we should work with robots once they do achieve intelligence etc. This is indeed what Asimov was doing by writing in the '50s.

adrian

These aren't robots in an asimov sense. They have no concept of harm even in the direct form of violence, let alone harm to society. The only sense in which they're taking jobs is in the form of mechanical replacement, which has been an issue since the start of the industrial age.

adrian

At what point would the use of AI be found to be in violation of Asimov's Rules for Robotics? When AI started to take away human jobs? That's already happening. https://www.britannica.com/topic/Three-Laws-of-Robotics

Sounds like there will be the need to put a tax on production profits, as well as import duties, such that the revenue from those taxes can be sent back into the general population to maintain goods and services necessary for standards of living, consumerism, and even human jobs.

I'm really worried about my daughter at this point :(

Stefano Landi

I think your concerns are justified. It's going too far... irreversible damage feels imminent at this point. Really scary.

Motten

what I worry about is the bad actors using malware to corrupt systems and a possible terminator scenario where lose control of the bots