Ep. 2 — Perry Metzger Makes the Case for Unrestricted AI Development

URL: https://share.snipd.com/episode/c2fc8425-0fb2-4d5c-b21b-4dfdefeb2778
Chad Hudson's AI podcast notes from Ep. 2: Perry Metzger Makes the Case for Unrestricted AI Development

In Fi, Episode 2. Welcome to EnFi, the future of finance
Key takeaways:

• This is episode 2 of the EnFi podcast hosted by Dr. Robert Murphy, featuring Perry Metzger as the guest.
• Perry Metzger is a technology manager, consultant, and academic researcher.
• He specializes in computer systems, networking issues, software development, system architecture, security, nanotechnology, and molecular manufacturing.
• The focus of today's episode is Perry's involvement in the debate over artificial intelligence.
• Perry is affiliated with the Alliance for the Future.
  Transcript:
  Speaker 2
  In Fi, Episode 2. Welcome to EnFi, the future of finance, hosted by American economist and author Dr. Robert Murphy. Each week, tune in for dynamic discussions with business pioneers about emerging trends in finance, life insurance, asset management, technology, and more. Now let's talk the future of finance. Welcome back everybody to another episode of EnFi. Today our guest is going to be Perry Metzger, who has numerous accolades, reading from his LinkedIn account, says I am a technology manager, consultant, and academic researcher, Specializing in computer systems and networking issues, especially those connected to software development, system architecture, and security. I also have a strong side interest in nanotechnology and molecular manufacturing. However, what we're talking about specifically in today's episode is Perry's interest in the debate over artificial intelligence or AI. And specifically, Perry is involved with the Alliance for the Future, which is an organization (Time 0:00:00)

The Benefits and Harms of Pauses in Predicting AI's Ramifications
Key takeaways:

• Pauses in the context of regulatory capture may benefit those interested in slowing down competitors or improving the profitability of their own company.
• Attempting to predict the full ramifications and societal benefits of regulatory capture is unlikely to be successful.
• The potential harms of limiting access to AI include its use by dictatorial governments and military organizations.
  Transcript:
  Speaker 1
  On the one hand there's the question of what good do pauses do and the answer is they don't do any good because you're not going to make any progress in attempting to predict anything we're In trying to figure something out after a while and they do do some good for people who are interested in regulatory capture right. If you're interested in trying to slow down competitors or get yourself into a position where you can catch up with incumbents in a field, ways of me capping the incumbents are often Economically valuable to you and there's all sorts of regulatory capture reasons why you might want to impose a eye regulation in order to improve the profitability of your company But if you're talking about what the societal benefit would be for something like that there's not. You can wait six months you can wait 700 years you're not going to predict all of the ramifications you're not going to figure out how to do this differently or better or in a way that changes The way that it affects human beings not going to happen but there's also the other side of this which is what sorts of harms does it cause and the answer is twofold first of all I worry a Great deal about getting ourselves into a world where the only people who have access to AI are the worst possible people which is to say things, diptatorial, governments, military Uses of various kinds and there's going to be military uses of AI right. (Time 0:22:50)

The Concrete Benefits of Advanced AI
Key takeaways:

• There are real concrete ways that AI can help humanity beyond just fun applications.
• The number of minds and amount of effort that can be applied to problem-solving has historically limited economic growth.
• AI producing unemployment is based on the fallacy of a fixed amount of work.
  Transcript:
  Speaker 2
  So yeah so now just turning focusing more specifically on the benefits that it's not simply that oh yeah it's kind of fun that you can ask chat gbt to do a wrap in the style of m&m talking About my history paper like it's there are other real concrete ways that this can help humanity that we're just on the cusp of unleashing so yeah can you talk a bit about but yeah since
  Speaker 1
  The dawn of time what has been the major things slowing economic growth and it has always been the number of minds that you can apply to any given problem the amount of effort that you can Push put behind any problem that you need to solve our ancestors weren't stupid right they weren't different from us cognitively if you go back several thousand years the people that Lived then could have been taught anything that the people now know but the population was vastly smaller and the pool of available information was smaller and the amount of labor intellectual Labor that could be applied to things was vastly smaller and there there are people who talk a lot about AI producing unemployment which is a ridiculous idea it stems from the lump of Labor fallacy the idea that there's a fixed amount of work out there and that when you only have so much work available and if there are people already doing all of that work the new people (Time 0:28:20)

The Cost of Aluminum
Key takeaways:

• AI can be used to hire teams and build new consumer electronics
• AI can drive down costs and make services more available
• When costs are reduced, people start doing things they never imagined before
• Driving down costs can open up new markets
• Expensive items can become more accessible if their costs are reduced
  Transcript:
  Speaker 1
  Dream that you have you can get enough labor to apply to it right you want to you know you think to yourself I really really hate the way this piece of consumer electronics works I wish it Worked another way well now you'll be able to hire huge teams of ais to work to build the new one that does the things you would prefer and programmers to write the software for it and people Often crudely think oh well you know if we have AI electrical engineers and we have AI medical researchers and we have AI attorneys and we have all these other ais there won't be any work For people but really what happens is that when you drive down the costs of many of these services and make things more available suddenly people start doing things that they never would Have imagined doing before if aluminum is more costly than gold then you'll end up in a situation as we did in the early 19th century where Napoleon served his most honored guests on aluminum Tableware and the reason he used aluminum tableware was because it was incredibly expensive but you make aluminum really cheap and you don't illuminate the market for aluminum the Five people who are buying aluminum before don't (Time 0:31:34)

The value of intellectual leverage in productivity
Key takeaways:

• The speaker admires the economist's collection of essays and wishes they had more time to delve into it
• Outsourcing certain tasks can increase productivity and allow for more publishing opportunities
• AI systems can amplify productivity rather than replacing human effort
• Many important inventions throughout history involve intellectual leverage, such as the invention of writing as a means of storing and conveying knowledge
  Transcript:
  Speaker 2
  Well said as you resonated just the other day somebody was asking I had a guest for a different podcast and an economist on talking about this collection of essays that he put in someone Emailed me and was asking me about it and I said oh well I just wanted to let the world know because this guy's work I've often said if I could clone myself the area that this guy's publishing Papers and that's where I would but I just I haven't had the time to get it deep in the that's what I'm saying that here but it's yeah if I could outsource a lot and the same thing to like people Who know like like professors who have research assistants like that's what it is is the kind of thing where they supplement what you're doing so certain things that you can outsource They go ahead and do it so you end up publishing more so yeah here it's the email G.I.U.s for people who watch the Marvel movies is like Tony Stark in his house creating stuff he's got his AI system that go do this let me okay show me the molecule and rotate it and he just it's not that the machine puts Tony Stark out of business it's that it amplifies and makes him able to make Some a thousand times more productive than he would be even using a computer but what didn't have a little AI personality helping him yeah well so all so many of the most important inventions
  Speaker 1
  In human history have been sorts of intellectual leverage of various kinds the invention of writing it was originally invented just to keep business records right I mean the earliest Cuneiform tablets we see where people were recording things like Joe bought five and fours of olive well from Fred at this cost or what have you they probably weren't named Joe and Fred No they were they probably had names more like well nussier or what have you but anyway the thing is though that people very quickly discovered you could store knowledge by writing you Could convey knowledge to other people you could avoid (Time 0:35:35)

Tucker Carlson's Stance on Self-Driving Vehicles and Job Loss
Key takeaways:

• Tucker Carlson expressed support for self-driving cars/trucks due to the potential for creating middle-class jobs
• The speaker disagrees with Tucker's view and highlights the potential benefits of self-driving technology, such as reducing traffic fatalities
• The speaker questions Tucker's stance by comparing it to not banning farm equipment, which could also create jobs
  Transcript:
  Speaker 2
  Yeah I like one just one specific example that I had heard you on other shows talk about different ones even beyond the what you just mentioned here in passing but I don't know if you saw This period but Tucker Carlson was being interviewed and someone asked him about self-driving cars or trucks and he said oh I would allow those in a heartbeat think of all the good middle-class Jobs that that would be thrown out you know people throw out and to me it was like are you out of your mind that for one thing if you know once technology gets good how many fewer traffic fatalities
  Speaker 1
  There will be so there's a bunch of yeah well they the 30 to 50 yeah the 30 to 50 thousand people a year whose lives will be safe we don't have to think about them yeah of course we could go further Than this why doesn't Tucker want to ban farm equipment because imagine all the jobs we would have if people had to hand harvest grain right good middle-class jobs working out in the Hot sun with a manual scythe (Time 0:38:50)

The risks of overregulating AI and the need for responsible use
Key takeaways:

• There is no reason to worry about the idea of AI wiping out all life or the human race suddenly vanishing
• Worrying about over-regulating AI could result in only the worst people having access to AI for malicious purposes
• The key to preventing misuse of super AI is having good people with access to it
• Comparisons to gun control debates and geopolitical issues highlight the importance of responsible use of powerful technologies
  Transcript:
  Speaker 1
  In all life being wiped out or the human race suddenly vanishing as the AIs decide to change the to turn the entire planet into paper clips paper clips is a common example people use in These discussions it's an incoherent possession and if you look at the details of the arguments that are made about this they are both at a high level and in detail almost all riddled With errors there's no particular reason to worry about it there is reason to worry however that if we decide to over regulate the technology early in its existence that we could end Up in a situation where only the worst people have access to AI and use it for terrible things that is legitimately a thing that could happen right if we decide that that we are so worried About developing AI that we're going to leave it only in the hands of sociopathic dictators and dictatorial regimes then i can guarantee yes there will be bad results because people Will not be able to use the technology for good and people who want to use it for ill will have no check on their bad behavior the only thing that can stop a bad guy with super ai is a good guy With super ai and well i mean what stops bad guys with armies that's what i'm saying like i'm rough and i'm like let's talk to Vladimir Putin from rolling over Ukraine right the gun control Debates (Time 0:56:09)

Wrapping Up the Discussion on AI and Podcasts
Key takeaways:

• The use of AI can automate tasks like listening to lengthy podcast discussions.
• The guest has a startup called Silent which focuses on improving software quality using AI.
• The guest is also involved in a nonprofit organization called the Alliance for the Future.
  Transcript:
  Speaker 2
  Yeah well putts so i think that's probably a good point for us to wrap this discussion up uh so we don't want to run for four hours i understand lex pigment does you know when we get the ai
  Speaker 1
  That they can just hand it out they can just copy our styles and then see where the conversation is gone the way you deal with that appropriately is to have the ai that listens to the podcast So you don't have to suffer through four or five hours of it it can listen to it on your behalf and then you don't have to well folks my guess has been perry mexger perry where can people go
  Speaker 2
  I know you have some organizations you're affiliated with the pernancy this discussion well i've got a startup which is not entirely pertinent to the discussion even though we're
  Speaker 1
  Doing ai research it's called silent i guess why leon i see and we don't really have a website because none of the stuff we are doing has been publicly released but we're working on using Ai to improve software quality by doing mathematical proofs of the correctness of software so five hour discussion all on its own and i'm also have helped with some other people have Recently started a nonprofit that should have a website soon we have not quite officially launched should have happened over the labor day weekend didn't quite call the alliance for The future and that can be found at affuture.org but it cannot be it'll be found in a few days not today and probably not even today the day you release this (Time 1:00:20)