AI's Fork in the Road: Societal Bliss or Existential Threat?
24:54 Share

Auto s best of twenty twenty four picks are here. Discover the years top audio books, podcasts and originals in all your favorite genres, from members and signified to mysteries and thrillers. Audibles curated list in every category is the best way to hear twenty four best in audio entertainment like a stunning new forecast production of George or wells one thousand nine hundred eighty four heart felt memories like supreme court justice ki Brown Jackson, lovely one the year's best fiction like the women by Christian hana and possible ever its brilliantly SUV sive James possible there is more to imagine when you listen, go to audible outcome slash, imagine and discover all the year's best waiting for you.

Welcome the money for the rest of us. This is a personal financial on money, how IT works, how to invest IT and how to live without worrying about IT. I'm your host, David stein.

Today is episode four fifty seven. It's titled a is fork in the road, societal bliss or existence al threat. One year ago, OpenAI released ChatGPT. When I first came out, I was pretty sceptical and then have begun to use IT.

And in the really a fascinating example of how a tool can evolve or be adopted to increase one's personal productivity, we first discussed ChatGPT three last january and upset four seventeen, will A I replace your jobs? And we talked about what generative A I is and how, basically, it's a very sophisticated word, predict, been trained on reams and rims and billions of dreams of data. And I was skeptical.

I wasn't using IT, was fascinated by IT. But, and we did a few experiments, but IT was okay. Six months later, july twenty twenty three, we released episode thirty nine.

By then I had OpenAI, had released ChatGPT four, which was much more sophisticated, had been trained on much more data. And I used IT. I had, as I mentioned, that that episode, some barely complicated, complex back and forth discussions with ChatGPT.

And I was at that time or shortly thereafter, that I invested in a couple of etf that I thought would be well positioned to benefit from A I in the years ahead. Now another five months have passed, and I use ChatGPT every single day on on a wide variety of task bread. Ask me to to oran a query regarding how advice on on negotiating for A A hand pan, which is a relatively new, fascinating instrumented place.

I provide ingredients to ChatGPT and IT creates recipe. I used IT for a cranberry sauce recipe. I've used IT to do translations, helping with checking spanish translations, understanding greek words, ask health questions to IT, mostly been using IT in finance.

We're releasing a video this week where the the formula is the merton share, which is a formula for figuring out how much to allocate to risk your assets. We discussed that in a podcast episode last month, but it's easier just to have ChatGPT put the formula there because it's it's already format IT and use that in the video. We've used IT for a podcast.

Outlines or clarifications helps with creating titles. We've gone back and forth and discussions, fairly complex discussions, underlying math of bonds and fixed income, including duration, yield, maturity, convexity. We've used IT to write ad copy. IT is some of our sponsors sometimes want us to write the copy will in iraq with ChatGPT to create the copy.

We ve used the dally version by OpenAI, which creates images to do youtube some nails that is successful at that little over the top and di cannot spell we have used this to summarized books and at this point its my go to source with any question I have on any topic including history, economics, finance but I don't trust IT completely so I always verify I would not trust my life to this bot David wine burger is book everyday chaos in speaking of A I said we don't use these technologies because they're huge, connected and complex. We use them because they work and and that's why I use ChatGPT every day and pay for, uh, plus subscription for IT that is twenty dollars a month. Back a year ago, a lot of the discussion regarding ChatGPT was, well, as is this human life, can we trip IT up to make IT do crazy things? And now here is past, and i'm using IT as A A research tool to help me become more productive.

One of the research providers we've used as capital economics, they released a detailed report this past september on A I their view. And I agree with you that A I can substantially help the global economy, individuals and businesses become more productive. IT will perhaps be revolutionary in terms of how software is developed.

IT can lead to greater corporate profitability as companies are able to incorporate AI into their systems, their processes, to help individuals. People were productive. I know that I am more productive today with the help of A I then I was a year ago.

Not that it's replace what i've done. I don't worry about ChatGPT or other A I taking over my job because as they say, I don't trust ChatGPT. And for looking basis, IT doesn't do a very good job understanding what's going on now because, for example, right now, the training data only goes through April twenty, twenty three.

And one of the things that you find with investing in finance and the economy is there's time decay. Things are always changing, we call IT investing on the leading edge of the present. And so I believe in most cases, A I is not necessarily going to replace jobs.

There are some translating languages that could be tough, but in in most cases is helping individuals be more productive in the job will create other job opportunities. So that's all happy news. And then we see what happened.

In the past couple of weeks. The C E, O, as OpenAI sam altman n was fired by the board unexpectedly. Ly, and then there was sort of a rebellion within the company among some of the investors to wear.

Now sam altman en was reinstated as the CEO. Greg brock, man who was on the board and and terminated, was also reinstated as the company's president. Open eye is an unusual structure, is that IT was established as a not for profit with aboard in place and then they've established a four profit enterprise that is taken investments from microsoft and other sters.

And is that for profit entity that releasing these tools like ChatGPT and di as part of this reorganization? With the termination reinstatement, there was a change in the board. Three members were pushed out. L, A suitable ver OpenAI, chief scientist, originally voted to to fire sam alt men, and then had changed his mind.

And one in him back, Helen toner, who is director strategy at Georgetown university center for security, an emerging technology, he was removed from the board but had voted to terminate sam altman n, as did tash macleay, who's an entrepreneur researcher at the rand corporation. This is an incredibly unusual situation. You don't see a board fire their CEO, then have the the CEO reinstated in the board, basically got fired.

New board members now include large summer, the former treasury secretary bret Taylor, or who is the former coach ef executive at sales force. And then there is one remaining board member from the original board, adam d. Angelo, that has state he's the chief executive of cora.

So the the board no longer has representatives from the what we could call the traditional AI community. A I arose out of academia, academic researchers. They over the years typically communicate via published research papers just the way other professors do.

And so I was sort of this academic research oriented type endeavor that is exploded in popularity. And now it's the hottest thing in technology. And then so there's a real contradiction there.

David Brooks visited the colonists at york times, visited the headquarters of OpenAI and interviewed a lot of people and really captured sort of the moment we were at. He writes, the people in A, I seem to be experiencing radical, different brain states all at once. I found that incredibly hard to write about A I, because IT is literally a noble, whether this technology is leading us into heaven or hell.

And so my attitude about IT shifts with my mood, financial times calling on this benedet events, you can spend weeks of your life watching videos, and machine learning scientist arguing about this and conclude only that they didn't know either whether this is a good thing or a bad thing. ChatGPT might scale all the way to the terminator in five years, or in five decades, or might not. And he equates IT to the ninety eighties with a newly invented byplay e.

And worrying about IT, whether IT would shoot in to orbit, we just don't know. And that's what so fascinating about the current moment. Here we have this tool that I use every day find IT incredibly helpful. And yet these harsh stories of things could go wrong in one of the things I wanted to do in my research this week, because I understand what are people worried about, like what exactly could go wrong.

And that LED me to something called effective altering ism, which is a research emphasis in a community that I first heard of in relation to sam, a bank man fried the convicted felon against this point of f. tx. egypto.

Currency exchange. And he was an advocate of effective altruism, apparently effective tourism. M is a number of board members at OpenAI are part of this movement. And I knew very little about IT. IT was formalized, I think, about ten years ago by some scholars at oxford university.

And the whole point is to figure out, within helping people in countries, what is that? That works, something as effective as giving away malaria nets so that individuals in less developed countries aren't exposed to malaria. The effective altering ism isn't so much defined by the different projects, although there are some surprising initiatives that they work on.

If you look at their website, they talk about more of of what their principles are, terms of understanding what works when IT comes to giving and research. And one priority is what they call prioritization. IT isn't just about doing good, but it's figuring out what ways to be good to give the help has the most impact and is part of that.

They value what they call impartial altruism. It's nice to to help about your family and friends or country. But again, the focuses on how can we help the most people and have the biggest impact.

They value open truth seeking, figuring out what is IT that actually works in using data to to determine that. And they value having a collaboration spirit, building community about being a good citizen. They write, working toward a Better world.

And so that all seems really good. And then you look at, well, what are these areas of focus that they believe will have the biggest impact? The first is global health well being.

And that make sense, will give example of malaria net or or other things related to that includes pandemic distance or or protecting future pandemic. The second area is not one I would have thought of, but apparently one of the original founders of effective altruism is Peter singer, who's been a leading advocate for animal rights. And animal welfare is one of the focus areas for effective altruism.

But the third, which is sort of the focus on on our topic, is reducing global catastrophic risks. And one of the biggest risk they see is missiles, artificial intelligence. Effective altruism has A A website called eighty thousand hours, which is A A site to help individuals figure out what to do professionally, which jobs will have the greatest impact from an effective altero ism filter.

Here is the top careers, the first A I safety technical research. They said the development of A I could transform society, and this job was helping crease a chance is positive, while preventing ways that A I could Carry out dangerous behavior. The second leading recommendation for a job have the greatest impact is A I governance in ordination.

The third is bia risk research strategy and policy. Fourth, information security in high impact areas, including A I and biotech y whether this technical ability is stolen. And the fifth is an expert in A I hardware so very much and focus on on A I and its potential threats and protecting against that.

David wine burger writes that machine learning systems, A I they're none moral. They're just machines. And if we just outsource morality decisions to a ion check, that is a statistical engine that basically could inflict harm on individuals.

And so we need what's called A I alignment, making sure that A I does things actually to human values and human intent. But there's still value in A I IT can actually help humans make more fair, more impartial decisions. Before we continue, let me pause and share some words from this week sponsors.

Before we continue, let me pause and share some words from one of this week sponsors. Net sweet, what does the future hold for your business? Ask nine experts and you'll get ten answers. Bull market, bear market rate will rise or fall, inflation upper down. Be great if we had a Crystal ball, but we don't.

Until then, over thirty eight thousand businesses have future proved their business with net sweet by oracle, the number one cloud E R P, bringing accounting, financial management, inventory, hr into one fluid platform with one unified business management street. There's one source of truth giving you the visibility and control you need to make quick decisions with real time insights and forecasting. You're peering into the future with actionable data.

When you're closing the and days, not weeks, you're spending less time looking backwards and more time on what's next. I know as our business growth will certainly consider using next week by oracle. Now whether your company is aren't millions or even hundreds of millions, next week helps you respond to immediate chAllenges and seize your biggest opportunities.

Speaking of opportunity, download the CFO s guide to A I A machine learning at net sweet that comes less. David, the guide is free to you at net sweet dot. Come slash David, net sweet dot.

Come slash David. Knowing where our one's going in, how it's being spent, helps us feel more at peace. Monarch money is a Operated all in one personal finance that gives you a comprehensive view of all your accounts, investments, transactions.

In more, you create custom budgets like we do, can track your progress toward financial goals and collaborate with your partner and now get an extended thirty day free trial. When you go to monarch money dot comes last, David. Monarch is the most customizable budgeting APP.

You can change the layout of your dash port. And I have done that, can talk, go between light and dark mode. I haven't done that, but you can set up automatic rules for transaction.

So you, I haven't to fix IT each and every time. And best of all, monarch never sells your data to third parties or shows your ads. After trying monarch myself, I understand why it's a Operated personal finance APP.

And right now, listening ers of this show will get an extended thirty day free trial when you go to monarch money that comes less. David, that's M O N A R C H M O N E Y dog comes last, David, for your extended thirty day free trial. Historically, there's always been a risk of what if, what if a ee takes over? Nick, boston philosopher, university of oxford, gave the example.

If, let's say, we developed is super AI, that could could make paper clipsed and that the A I engine decided that killing all humans was pivotal to its mission of creating paper clips because, well, humans could turn the machine off, or they have atoms that could be used to make more paper clipsed. And know that that's clearly an absurd situation. But that sort of an example, what A I researchers that are worried bad to think about, the idea that if we don't specify what A I systems can do, that they might do unexpected things.

I ask ChatGPT, because this is a tool I use everyday, how my large language models like ChatGPT pose an existent al threat to humans. I wanted to understand what the scenario would be. And IT pointed out several areas.

One, IT could just get advanced capabilities that as A I continues to evolve, IT could gain more sophisticated abilities, become super and surpass human intelligence, come uncontrollable or manipulative. Now we're so far from that. And and one of the things in researching this episode, they have to be trained on data.

And ChatGPT four was trained on on so much data, but in some regards, the running out of data to train the systems on. So the biggest worries that, well, IT just gets way smarter than us and IT can make decisions without human intervention, and that those decisions could not alive with human values. And and as a follow question, well, I needed some examples to be reminded of examples where is A I currently making decisions without human intervention.

And IT gave examples of autonomous vehicles, self driving cars and and there are have been issues with self driving cars recently, and and some of the self driving car companies have had to pull back because of accidents as the car. The I made decisions that harmed people. But that's one example that that's currently happening.

Financial trading algorithms, trade are being made without human innoventions by AI assistance. Potentially some medical diagnosis tools are making decisions, are interpreting data such as for x rays and m ize, without human intervention. Now this a very specific case, but this is an example where in many cases, I just can do a Better job than human.

And looking at an x ray for things that that a human I might not pick up. A, I makes autonomous decisions when IT comes to energy management, so managing the baLance of a supplied demand for electric networks, A I is making content moderation decisions on social al media platforms. ChatGPT mentioned that A I is being used for predictive maintenance, making requests for when certain maintenance needs to be done, dual repairs.

There is a whole idea of autonomous weapons systems where drones, A I drones, are making decisions of who, where and went to attack. Or at least it's being discussed if it's not happening now. Retail mentor management predicting inventory needs and going ahead and ordering a replenishment of that inventory.

And then I mention smart home devices, trolling, lighting, heating, security. So so it's already happening. And I ss, these systems are doing that. IT still doesn't seem like an existential threat yet.

But I wanted to know all right how there's this concept called the singularity, where A I becomes so advanced that IT undergoes a rapid self directed improvement, reaching a point beyond human control or understanding. IT becomes smarter than us and starts to act on its own. And this singularity run away.

A I bats a potential threat and existence al threat. But I wanted to know how how could that happen? I guess there's a little alarming because I has ChatGPT this and IT broke out five phases.

Phase one is accelerated AI development research to develop a new type of machine learning algorithm that significantly surpasses the capability of existing systems. It's more efficient, requires less data, and can generalize across different domains, and so IT starts self improvement loop, improves its own algorithms. IT creates versions of itself that are increasingly more intelligent.

An efficient that face one, face two IT surpasses human intelligence IT far exceeds the bride's human minds in every field. IT conducts its own research and on development activities. IT could discover new scientific principles, new technologies that humans are not even able to understand, and at least to face three rapid and unpredictable developments.

IT improves that an expected entity rate. IT starts creating technologies and solutions at a pace and complexity that humans cannot keep up with or fully comprehend to wear human lose control of IT. Our ability to control AI actions diminish rapidly and then face for societal impact, and transformed society directly, indirectly, to where humans depend and heavily on aid in order to exist.

And then face five is is just uncertain outcomes. Positive scenario is IT works toward the veteran of humanity, solves major global chAllenges like poverty, disease, environmental degradation, an unprecedented era of prosperity. And the negative scenario is AI accident ways that are missing lined with human values and interest leading to unintended and potentially catastrophic consequences.

That's where we're at. But IT seems like we're pretty far from that. IT is like a byplay e it's an incredibly effective tool. But IT is IT is not yeah and leashed some of these dire scenario. But I guess what a little alarming is in reading what A I researchers have said IT there's a real diccon's y there.

Some are incredibly including apparently some of the board members that open a ee that felt that sam altman and his team were moving too quickly and not being cautious enough, but others are less concerned about that. I think there's time to make adjustments. Been one year now, A I has been coming along for for decades.

But these large language models ChatGPT being A A A prime example that new in terms of their power, but is still very, very, very early. It's hard to regulate something if we don't know what we're regulating against, what are we protecting against? Now I guess they can put some frameworks in place.

But because it's so early, it's hard to to say we don't necessarily want to outlaw IT because I can significantly help our productivity and create things that haven't been created before working with humans. In conclusion, I I think where we are is U Z. I figure out how I can help you in your work, in your leisure.

But it's not human. It's a tool that can help us be more creative. I suppose IT can be misused.

why? No IT can be missed. But I don't want A I to replace me as a podcast because IT would be boring.

There is no story to IT, no human element to IT. IT can help with research, but ultimately the people listen to podcast because it's it's a narrative medium. There's a motion being conveyed, reaction interviews in some cases.

So experiment with IT and perhaps do some investing in the space. I i've taken a couple positions in, in two growth or entity tf. They mayor may not work out.

I've mentioned in that upset and july that if productivity increases, corporate profits increase, that will benefit the entire global stock market and they'll be growth that way, assuming some of these more alarming scenario don't come about. And that's why we diverse. So that's my up data on where we stand with A I.

We are at a fork in the road, but still very early days. I think there's time as A I developed to put in more regulations or other governors in place. So we do get this A I alignment so that A I is making decisions consistent with human values.

It's it's too early to just shut down. It's just like with anything when IT comes to a retirement investing, for example, career, we can make adjustments over time. We can be flexible, nimble and adapt as things evolve.

The worst thing to do is just not be involved at all. And so it's sort of all these developments are happening. We're just not ware of them are participating in them.

It's Better to participate so that we can help guide the process. That episode for fifty seven. Thanks for listening.

I have loved teaching you about investing on this podcast for over nine years. Some topics, though, I just Better explained in writing or with a chart. And that's why we have a weekly free email newsletter, the insider guide. In that newsletter, I share charts, graphs and other materials that can help you Better understand investing. It's some of the most important writing I do each week. I spent a couple hours on that newsletter on wednesday morning as I tried to share something that will be helpful to you if you're not on the list, please subscribe, go to money for the rest of us dot com to subscribe to the free insiders guide weekly email news letter everything I have shared with you in this episode ment for general education, i'm not considered your specific risk situation, have not provided investment advice, this is simply general education on money investing in economy. Have a great week.