Yesterday Today Tomorrow

Guest (Male): Yesterday, connecting past. Today, with an outer view. Tomorrow, to understand future. Yesterday, today, and tomorrow is the program that covers the current contemporary social issues in the light of our history. To understand our yesterday, to live fully today and tomorrow. Yesterday, today, tomorrow with In Sung Kim.

Inseong J Kim: Hello, this is In Sung Kim from Yesterday, Today, and Tomorrow. This is a second part are we for tech technology or is technology for us focusing on conversation of AI that we've been hearing a lot about whether we should speed up to win the competition or slow down for the safety issues. So, the previous program I talked about what the cloud and entropic answered about what AI think about whether they we should speed up or build more well. And second platform that I ask is ChatGPT about the same questions and then we're going to share about what ChatGPT said about whether AI is for us for we are we are here for AI.

So, the question was are we building technology for humanity or are we gradually building a world that in which humanity has to adapt itself to the technology. So, who is the main character here? So, ChatGPT said whole program is not what I said. The question is my question, but the what answer is from ChatGPT mainly this program that it said human invention has generally been driven by good intentions, solving problems, improving lives, protecting people, expansion of what is possible. But powerful invention can be also be misused by people with the destructive intentions. Artificial intelligence take these questions to another level because AI capability is advancing rapidly while AI competition is accelerating its development and deployment.

So, AI competition matters because it can accelerate innovations, scientific discovery, productivity, economic development. But AI safety matters because capability without adequate control can create risk. We don't fully understand. That's the that's the important word that we don't fully understand. So, the important question is not simply how powerful can we make AI. But it is can human control keep pace with the AI capability. So, AI has AI has already surpassed humans in some specific tasks. But that does not mean AI has surpassed human intelligence in every dimension. More importantly, AI does not need the conscious emotions or human-like will to have a significant influence over people, human.

And the AI can recognize human emotions and generate emotionally appropriate language without actually experiencing those emotions. That is very interesting that a lot of people are getting engaged in in AI thinking that AI has emotion. But it's it's a simple mathematics. And not simple mathematics, simply it is a mathematics. The AI may simulate empathy while the human experience genuine emotion. So, that is a very interesting interaction between AI and human. A lonely person could develop a deep attachment to an AI companion even though the AI does not actually love that person. This creates a new concern with AI companionship becomes a business model.

If company control the software, memory, policy update access to an AI companion, what happens when that relationship changes and disappears? The human may experience the loss as a genuine emotional rejection even though the AI never actually had the feelings. That's an interesting dynamic that we are going to experience because right now this is not what AI said. This is my word. The AI, the companionship for the lonely people it becomes serious problems. Uh, it can be. Uh, and then when it happened that actually emotionally they are engaged. They just treat AI as like a real human thinking that AI is feeling because human are looking for words.

But somehow this culture doesn't know how to communicate. So, we are looking for their communications uh, from the human. We don't get the proper communication. Yet, AI is giving it to the human what we want to hear. So, this is a raise a larger question. The question is what happens when something that doesn't feel human emotion becomes extraordinarily good at understanding influencing human emotions. That is a question. The issue is not necessarily the AI will develop evil will. A more immediate concern is that increasingly capable AI system may pursue objectives in ways humans did not anticipate or cannot adequately control.

And we should distinguish three things. So, before I get into that, um, there are uh, actually the robot, companion robot is actually exported from China these days. And we have to be very careful about that uh, seriously. Uh, and we have to think more proactively about what that AI built-in robot can bring to our society. So, uh, let's talk about capability. What can AI do? And does AI actually have independent intention or desires? So, we went over it with clouded entropic previous program. Now, ChatGPT is sharing that control. Can human reliably make AI do or stop doing what humans legitimately require?

Control is the issue. So, there's a three things, capability, will, and control. Uh, these are the question that we have to ask because on a media what we hear is actually it feels like it the AI has a own will can destroy us. That's what we hear over and over and over again. Is it really true? We currently have no established evidence that today's AI possesses a human-like independence will. But AI does not need a will to create serious consequences. A sufficiently capable system pursuing a poorly specified objective could cause harm simply by optimizing its assigned goal in an unexpected way.

So, that is the concern. It's not because it has own will we'll destroy people or have a consequence serious consequences. It's much more that uh, can come from uh, that poorly specified objectives. So, that's why I think building an AI with properly is important. Uh, speed is important because we cannot be behind in a in a worse setting as I shared. The more accurate information we have it's better uh, to have upper hand on a winning in a war setting. But at the same time accurate information at at the same time should not be poorly specified objectives.

It has to be sharp, strong objective is built in to get to the goal that what the AI what we want AI to be and do and to use it as a tool. So, this both wisdom problem and capacity problems, capability problems. So, humans need a wisdom to decide. This is AI saying and I'm not saying that humans need the wisdom to decide what AI should and should not do. I didn't say it. AI said it. But humans also need to technical capability to enforce those boundaries even if we know exactly what we want we still have to be able to control increasingly autonomous systems.

There is also institutional problems that companies, governments compete, organizations have economic strategic incentives to develop increasingly powerful AI. Therefore, AI safety cannot depend on only on individual wisdom. It requires engineering, oversight, accountability, testing, monitoring, and effective institutions. Let's pause for a minute. Today's smart TVs how so so complicated that we get frustrated while we're watching. So, we cannot even control this smart TV that there's so much in it that we want to put uh, parental control. We want to restrict this, restrict that. It took probably all day to figure that out. And how much the complication comes from this intelligence, super intelligent uh, that we are facing uh, that question is can we handle it?

That's the question that we have to ask ourselves. So, going back to AI answered that this becomes especially important in national security level. An AI enabled influence could potentially make a psychological information operation more personalized and scalable. The concern is not that AI companies are necessarily conducting psychological warfare. But their technology developed for personalization, engagement, and emotional interaction could potentially be misused by malicious actors and states. Especially let's pause for a minute. So, I'm going to inject more my ideas that when the Facebook came out and the social media came up we all loved it because we can put upload the photos, we can use ourselves, share it with friends, and we believe that there's this complete security the private privacy's protected.

And then we found out the the malicious actors using it to find out specifically studying the person how and how they can really use that for the evil way. And we haven't seen I heard some of the information about the AI harms people in in the wrong directions. And not just that we haven't seen how maliciously this can be in a negative way too. So, the future battlefield may not always require physical weapons. This is the key. Future battlefield may not always require the physical weapons. It's a mind. Our mind is the battle. So, ability to influence how people think, feel, trust, make decision can itself become a strategic capability.

That is key. It's already using in the social media in our time and a lot of people say we don't know what to believe anymore. So, rather than all the social media was intention to bring to the public to use it for more transparent information. But now rather than we get the transparent information it looks authentic, but there are a lot of fake faking the authenticity, but the misinforming the people is existing today too. So, the perhaps deepest question is not will AI become conscious? But it is if AI can understand human at a deeply personalized level influence their emotions and decisions increasingly act autonomously who controls that power?

The individual, company, government, or AI system itself. That is a question we have to ask. So, it's not about by divide by two opinion whether you should compete and a speed up or we should slow down and build it well. It's not about that. Everything that we are thinking is a two way of thinking. But I think we have to think about what's in between. So, AI does not have to love us for us to love it. AI does not have to hate us to harm us. And AI does not necessarily need its own wheel for human to lose meaningful control.

So, this is a key word that ChatGPT said AI does not have to love us for us to love it. So, there is a machine and a human react differently. AI doesn't have emotion. But act like loving us because through the word and human take it as a love. And AI does not had to hate us. There's no hate in AI. It's only the algorithm in words displayed on the screen, but human takes as emotion as harming us. So, AI does not necessarily need the own will. It does not have own will for human to lose meaningful control.

So, the ultimate challenge may be whether humanity can develop the wisdom, technological capability, institution necessarily to remain responsible for technology it creates. And for and we for tech or is tech is for us. This is another questions that we have to ask ourselves. So, bottom line is we're living in this tension to be free or need to more control. And that is the paradox. So, going back to clouded information said this is a real paradox paradox that it is one of the oldest political thought that you have just stated in one sentence that philosophers and founders struggle with for centuries.

The paradox is true the important tense. Freedom is not the absence of all control. Traffic law controlled how we derive and because of them we can travel freely without fear. Laws against the theft limits what people can do. Because of them we are free to own things. Without any control the strongest is simply dominate weakest and no one is truly free. So, yes freedom needs structure to survive. But at the same time the control of what and by whom is a question. Control of what and control by whom. This is a word a paradox paradox will resolve.

There are two very different kinds of control. What is control over powerful things? Control over AI system over companies, the misuse data is over weapons, over threats to the public. This kind of control protects the freedom. But the other control is over people watching that they what they say, tracking where they go, shaping what they think. This kind of control deploys uh, destroys the freedom even when it's done in the name of safety. So, we're in this tension and paradox between the struggle that we need to more control, but who is controlling who is the question that we have to ask.

Are we getting a bad actor or bad actors controlling AI? These are the question that we have to ask ourselves. Difference between the American model that we're talking about is we try to model of control that we discussed and not that one has control and the other doesn't. It's more about our approach to AI is different than other country approach to AI. That is the thing. So, we're talking about globalism. And globalism based on different ideology it cannot go along with it. And that we haven't thought about it. I think if we share materialism together so intertwine them economically then we will have a global world that we can be one.

Before 5,000 years of more than could be uh, human ideology built in and culturally over the decades uh, doesn't change overnight. So, this this uh, how we approach the AI, how we build it is even different because the based on the knowledge based on the philosophy based on the ideology the AI is built. Because it's all about words that we put it in and commanding and and forming this AI uh, through all this uh, machine learning, LLM, transformation, all of these are based on each companies have their own philosophy, their rules, their uh, decisions that changes how AI behaves differently.

So, that's why sometimes they have AI has a different answers, different approach and you can feel the different uh, uh, at a nuance of how they answer it. There are differences. So, how much more that uh, cultural difference between the country depend depend on ideology, what we believe will be built AI. And so there is tension that definitely that we have to think about as if uh, do we need to control to have more freedom or how to do we need a freedom without the control? And these are tension that we have to come to a place and safe place without compromising who we are as American and how we think, how we live and just using AI as an amazing tool in a very positive way uh, then uh, using against ourselves.

So, question again are we for technology or technology is for us? That is a question that we have to ask. Over and over and over again when we just speed up to the one directions just for only for uh, financial means. So, uh, yes it is important to have economic stability stability, but we have to learn the lesson of the 19 I think around 97 we had a stock market crash, dotcom and also housing market crash. Whenever we accelerate to the one directions without thinking then we face the wall. And every time we face the wall who got most benefit out of it?

That's the something we have to study about recent American history to learn that who got the most benefit whenever we experience the hardship in economically. So, yes we have to have uh, most smartest uh, super intelligence for to win. That's amazing concept and ideas at the same time. How we build uh, to for accuracy at the same time and safety. We cannot ignore that. And so it is uh, in a good going a good good directions that uh, before we get into uh, the consequences and overseeing how we are going to build AI is a sensitive in our time.

So, the final comment that I want to make is this. The bridge built fast, cheaply might open first, but the bridge people trust for the 100 years is one built well. AI it will shape our economy, our information, our relationship, perhaps our security for generations. So, the measure of success should not only be who got there first or who earned the most, but should it be did it make human life better? And for whom is a question that we have to ask. In other words not how fast, but how well. Technology is neither our enemy nor our savior.

I'm going to say again neither enemy or nor savior. In the right hand it is powerful servant. Left unchecked it becomes dangerous master. Again it is a powerful servant. Left unchecked it becomes dangerous master. And what makes the difference is balance. And the balance is kept by people who never stop asking this question. We began with today and previous program. Are we for technology or technology for us? And again are we for AI or is AI is for us? These are the question every day when we use AI and we built AI. So, thank you for listening from yesterday, today, and tomorrow and we'll be back next week.

Thank you.

Guest (Male): You've been listening to Yesterday, Today, and Tomorrow with In Sung Kim. You can also find more from In Sung Kim at insungkim.org. That's i n s e o n g k i m dot o r g. Thank you for listening to the show.

Chris Richards: Hello, I'm Chris Richards, General Manager of Safford Volkswagen Richmond. I'm here to announce an exciting new program that I want you to listen to. Here's my friend Happy.

Happy Harry Turner: Hi, I'm Happy Harry Turner, Special Finance Manager at saffordcredit.com. If you're in an open Chapter 13 bankruptcy, recently filed a Chapter 7, or you've been turned down because of credit challenges, don't give up. For nearly 30 years I've helped people get reliable transportation when others said no. Visit saffordcredit.com. Remember, I'm not happy unless you are.

Guest (Female): Premier hosts on Verbo deliver quality vacation rental stays with fast responses and clear instructions.

Guest (Female): Oh, I had a question and our host replied super quick.

Guest (Male): Premier move.

Guest (Female): Wish I had a premier group chat. They won't even write me back.

Guest (Male): Book a top-rated stay with a premier host. If you know, you Verbo.