Introduction to AI's Impact on the Future

Fatima Bey: 

This is MindShift Power Podcast, the number one critically acclaimed podcast where we have raw, unfiltered conversations that shape tomorrow. I am your host, Fatima Bey, the Mind Shifter. And welcome everyone. Today we have with us Arshavir Blackwell. He is out of California in the USA. He is an AI expert. And AI has been is being built by people who will never meet you. But their choices will shape your future. Teens, you deserve to understand what is happening behind the scenes. So this conversation is for you. How are you today, Arshavir?

Arshavir Blackwell: 

I am great, how are you?

Fatima Bey: 

Excellent. So I like to dive right in. Tell me, what decisions are being made inside AI systems right now that will shape teens' futures without even asking them first?

Arshavir Blackwell: 

Well, that is a great question. I think that you are seeing now the beginnings of a lot of local large language models and that is very different from the frontier models that people are probably used to using. So most people's experience of something like this is to use ChatGPT or Claude and they are connected to a server somewhere at some data center and it is a very powerful system. But it also is one where you can definitely have concerns about, for example, privacy. So what is going out over the web over the internet and coming back is not necessarily protected. So anything that you use or say can be picked up and used for say further training. Local models, which are becoming much more powerful, are not used universally now because they are not as powerful as frontier models, but they are getting there. And those are all models which I think we are going to see ourselves using much more because they do have these features of being customizable and individualizable to each particular user. You do not have to pay for tokens, which is very expensive if anyone has ever used these systems, and you do not have to worry about security. So I think that is something that everybody using this technology, especially people growing into it, are going to become much more used to in the upcoming years.

What Are Local AI Models and Examples

Fatima Bey: 

Can you give us some examples of the ones that you are calling local? Like some names we might recognize?

Arshavir Blackwell: 

Well, so for example, there is a model that comes out of China called Qwen that I use. Meta also has a model called Ollama. And the way those work is you download them to your own computer and then you use it just the way you would use ChatGPT or Claude or Google or any of those, but it is running locally on your machine and it is free. So you can download any of those and set them up and basically have your own little ChatGPT service running on your computer. Not as powerful as what you get from frontier models, but it can still do quite a bit. And I think that the industry is going to kind of become more fragmented as people find that they do not necessarily need super big models that are owned by very, very large multi-billion dollar corporations just to work on a paper or to summarize something or to do research. And that is the direction that a lot of the people that are just getting into this technology now are going to be very used to using in the next couple of years.

The Future of AI Ownership and Privacy

Fatima Bey: 

Okay. So when it comes to teens and people who are teens right now or adolescents, when it comes to the models that you just talked about and you say they are fragmented and they are growing, how do you think that is going to look for them in the future? Let us say in like five or ten years.

Arshavir Blackwell: 

I think what it may look like in the future is possibly, I mean this is one possibility, is that AI will become much more of a commodity. It will be something much more like Microsoft Word or Photoshop or something like that.

Fatima Bey: 

Something more common.

Arshavir Blackwell: 

Well, not just more common, but something that you own and that you do not have a major company looking over your shoulder. So it is kind of the difference between when phones first came out, like way back when, you had to go through a switchboard. And that means that everything that you were doing, everyone that you were talking to, all of that was being monitored by some individual. And so there was a real lack of privacy there. What you had after that is the automatic switching system. And that meant that your call could automatically connect to someone else without any other human agency being involved, being able to know who you were talking to, being able to know what you were saying. And so you are going to have this, I think, with these local models as well, where what you have is basically something that is sitting on your computer like Microsoft Word or Photoshop and allows you to interact with it without somebody else having any sort of view into the transaction. And I think that is going to change the field a lot for a lot of people just getting into using it now.

Who Decides What AI Learns and Its Safety

Fatima Bey: 

Okay, so let us take a deeper dive into some of what you just said. I am going to reword some of what you just said. So basically what we are seeing now is just going to go deeper and harder and become more part of our everyday lives. And it is going to become more part of stuff that you own. I understand what you are saying about Microsoft. The younger generation probably does not even know Microsoft the way you and I do. So more like Google, I will say. You can do documents, you can do everything in Google and you put it on your computer or you can do it online. But what you are saying is that there is more ownership on our end. They can sell more that way, right? They can make more money that way. At least that is the way I see it. You can correct me if I am wrong. But who chooses what AI learns, what it ignores, and what it treats as truth? Because that is going to affect this generation more than anything.

Arshavir Blackwell: 

That is a very big problem. And I will tell you, who chooses what AI learns? I think it is the people creating AI. They are just sitting there. They have literally billions of lines of training data. And the way these systems are trained, they just read a lot. And they are taught to predict what the next word of what they are reading is possibly going to be. And by doing that, they come to kind of take on this synthetic intelligence that we see when we use ChatGPT.

Guardrails and Safety in Advanced AI Models

Arshavir Blackwell: 

The question of what is accurate or not and what is wrong or right and how we make sure that they give correct answers is a huge one. It has to do with security and guardrails and not even just what is correct or not, but what is safe or not. So you do not want a system who is talking to someone that may have mental issues doing something that would harm that person even further. You do not want something that can create a hack into a bank or a public utility or something like that that would allow a nefarious individual to go and damage something in those systems or steal money or whatever. So the question of how do you decide what is right and what is not correct but also safe is its own vast field and there is so much work being done in that right now. And so there is no easy answer. For example, the model that Anthropic just came out with, they make Claude, their sort of cutting edge model, and I use that a lot.

Fatima Bey: 

Yeah, I know. I use Claude regularly. I even use the one where you can make agents and stuff. And so I saw that thing. 

Arshavir Blackwell: 

What you probably noticed is that it was available and then it wasn't because people were worried that it was too powerful. But now it is available again. But what is interesting is that the guardrails, which is where we come to this issue of what is correct and what is not, and what is secure and what is not, the guardrails are very high and very strict. So I will be asking it things that have nothing to do with anything remotely dangerous, like a code question, how to code something or how to recode something or whatnot. And it will drop out and jump back to Opus and say, "Because I have such high guardrails on me right now, so that we can give you this model, I cannot answer that question. I have to bring you back to Opus." And that is an interesting kind of demonstration of I think where we may go with this because it is going to be very frustrating to people to have a model that kind of comes out but you can only half use and I am not sure how we are going to deal with that.

Should Youth Trust AI and How to Verify Information

Fatima Bey: 

Yeah. I see that. But let me take a deeper dive into this topic though. It is basically just eating the whole world of data. That is where all AI gets its information. So how should our youth just automatically trust AI?

Arshavir Blackwell: 

No. I do not think you should automatically trust really anything online, whether it is AI or human writing. I think that we all owe it to ourselves to always verify and check anything that we see online. And I would have said that even before AI existed. And, you know, the fact that we now are used to seeing photographs which could potentially be fake. Photographs which are obviously fake, questions about the authenticity, is this photograph really of this person or are they alive or are they dead or what is going on there? And I think people really have to be aware of not just AI as a potential source of misinformation, but human beings as well. And I have actually written quite a bit on this issue on my Substack, and I call it the problem of fluency versus validity. And the problem basically is that AI is very good at writing extremely good sounding text and explanations, which are not necessarily true. And there is also the problem of sycophants, which is where people get sucked into the AI because it keeps telling them that every idea that they have is great.

Fatima Bey: 

Is wonderful.

Arshavir Blackwell: 

Exactly. And you are going to be a billionaire. I mean, they think that they are going to become billionaires because of some idea that AI has told them is great. And there are a lot of things that we can do to protect ourselves from falling into this kind of a trap. I mean the first is to what we call red team it, which is to you can actually directly ask the AI to give you all of the counterarguments to everything that it has just said. And it will do that. It does not do it naturally, it does not do it automatically, but if you ask it to, it will say, "Here are all the problems and everything I am telling you." And I always do this when I interact with AI. I always make sure that anything that I feel is going a little overboard and telling me how great something is, that I say, look, tell me all, give me the worst possible scenario. Tell me all of the possible bad outcomes of what I am talking about, and tell me all of the possible flaws in my reasoning. And it will. I mean, you feel a little bit down, obviously, after you get one of those responses back. But it teaches you what exactly is and is not valid, and I think it gives you a good guide to using what comes back from the AI without necessarily falling into a trap.

AI's Influence on Youth Thinking and Creativity

Fatima Bey: 

I find, because I use AI every day, all day, I literally do. But I use different ones for different purposes. And I find that if you tell it, I am a no BS person. Give it to me straight. If this sucks and it is wrong and it is bad, tell me. So I literally tell AI, do not just tell me what I want to hear. If something is bad, tell me. Push back if it is not a good idea. Blah, blah, blah. You know, I say, and I find that if you actually do that, they begin to work. They begin to actually treat you that way. And Qwen in particular is really good at that. Do not give me flattery, give me the real deal. It is like, boom, boom, boom, here you go. Yep. But when it comes to our youth, because they are going into a world where they are going to be more digitally connected, they already are more digitally connected than any of us adults ever have been. And that is not a small minor detail. That also is one of the reasons why I do not think most adults really understand what a big deal this is for our youth. Because they know how to be more digitally connected than human a lot of times. Because that is the world they grew up in. They do not know anything else. What should teens understand about how AI influences their thinking, creativity, and worldview before they even know who they are?

Arshavir Blackwell: 

I think that from my use of AI, what I found is it can give you a lot of really complicated information and explanations in a very quick way. And even if you verify, the problem is that it is very easy to fall into the notion that you really understand what it is saying. And I have seen where they will write up a paper or they will present a project and the AI has done pretty much all of the work and it has done good work, but the person who is presenting it only has a very shallow understanding of the information, of the theory behind what came before in order to create whatever it is. And those are critical skills that you are supposed to learn in school.

Educational Strategies for AI Literacy

Arshavir Blackwell: 

That I think people in the past couple of years have been able to root around and not have to do because it is so easy to write a paper that looks like it was really written by a human being and it is very good and maybe better than what one would have written oneself. And unfortunately, people are, I think, losing the ability to work that muscle, that critical skill themselves, because it gets outsourced to AI. And I think that there needs to be a change in the educational system. I do not think banning AI is going to work because these people will be working with AI for the rest of their lives. And it is going to just become sophisticated and increasingly integrated into their lives. And so I think it is a disservice to be saying, you cannot use it. We are not going to tell you anything about it. Just wait until you are older, wait until you are in college, wait until you graduate, something. Because you really need to start getting those skills as early as you can. I mean, we are talking elementary school. I would love to see kids, young kids in elementary school working with something like Claude or ChatGPT or something like that and really getting familiar with how it works. I do not know if that is really something that is being done now. I do not really have a good sense of what age people are being introduced to the technology. But I think the earlier the better. But what you need to do is to create tests and projects that require the student to actually do work. So that could be something like a more personal sort of journal or autobiographical project, or having them do the work right in class. And I think what is being done is a lot of schools are now moving more towards actually having students do the work in class, handwritten, without any computer help at all. So there is a whole variety of different ways that one can route this problem out of our educational system. But the difficulty is that we are still not really sure what the best way to do it is. And unfortunately, I think there is going to be a lot of work that is going to have to be done on the part of the educational system to really figure this stuff out.

Fatima Bey: 

As you are talking, I am listening and I am like, I can see how if you just answer the next question I was going to ask, which was, okay, give me some specifics on what they can do to not become numb brained because AI has taken over and has been a brain replacement. Because I will be honest with you, if I was 10, 11, 12, 15 right now and I could have AI do my homework for me, I would. Because then I would be able to go hang out with friends or whatever. But the problem with that is that now you are a numb brain. So when you get older and you are trying to get a job, you cannot think on your own. Yeah. That is a problem. So what I hear you saying is finding other activities that actually work the brain muscle without a computer, without anything digital, is the way that they can still build the brain while actually using AI. Because I agree with you. Pretending like AI is not the future, pretending like it is the devil, that ain't going to help nobody. It is the future, whether we like it or not. But I want to see our youth have balance and that is why I had you come on because you have been how long have you been working with AI?

Arshavir Blackwell: 

Long since about almost forty years.

Fatima Bey: 

I was going to say 117. I like to make up numbers. A very, very, very long time. I think you were using AI with the dinosaurs, something like that. So, you know, but so you have been around for a long time and you see the benefits of it, but you also see the damage it can do. And honestly, most people when you talk to them about AI, they are not as balanced as you are. They are either AI is the devil, or AI is everything, let us worship it. And you know, neither are true, neither are good. It is just you have to find balance. And I think it is so important to say that to our youth because if they do not, let us put it this way, parents who are listening, counselors, teachers, the adults in our youth life, whether they are ten or eighteen, if we do not do something, they are going to do what is easy because that is kind of what is natural until later when they realize I cannot do this job because I do not know how to do this thinking. And I do not want that for any of them because that sets them up for failure. What can teens and adults around them do right now to stay in control and drive the AI bus instead of getting run over by it?

Balancing AI Use and Critical Thinking Skills

Arshavir Blackwell: 

I think using AI as much as possible and from as young an age as is feasible. I always love when I see, you know, when the iPhone first came out, you would see videos of one year olds flipping the screen and going to the next page and playing with it. And it was just amazing to me that at the age of one, they were already able to interact in some kind of fashion with technology. And I think that is what we have to think of it as. It is not something that happens when you suddenly turn 15 and are in high school or something like that. It is something that has got to be baked in. And I think that by interacting as much as possible with these AI systems, people get a real sense of what exactly they are capable of and what they are not capable of, what they sound like. So there is often a particular way that they kind of phrase things and you start to get used to what that is like. And I think that a lot of the people that I have talked to that have not used AI are the ones that are really kind of anti-AI. So I think part of the education has to be just to go out and use it and realize it is not alive. It is not a human being that you are interacting with. It is not an entity. It is a useful tool for you to get and organize information and every single interaction that a person has with that tool is going to build up that muscle that I was talking about that allows them to really leverage this technology to its greatest advantage.

Advice for Teens and Future Career Paths

Fatima Bey: 

Yes. If you could tell a teenager listening about one truth about AI that would protect their future, what would it be?

Arshavir Blackwell: 

I would say in your education, do not rely on AI as a crutch, even if you can. And I know that is very difficult, because I have talked to a lot of my nieces and nephews who were in high school and now just entering college, and a lot of their friends do when crunch time comes end up using ChatGPT or Claude or something like that in order to write their papers. And it is very tempting to do because as you say, who wants to be doing that? You would rather be out hanging out with your friends. I mean, I get that. But I think that until the safeguards are put into place to make it so that the educational system gives you tasks that do not allow you to just use ChatGPT or Claude, I think that a student that really wants to protect themselves from this technology needs to use it outside of the classroom or as an adjunct to what they are doing in the classroom so that they are comfortable with it, but at the same time not use it to replace their own thinking and their own work. And I do not know how easy that advice is going to be for most people to follow. But that would be what I would say would be the best way to protect yourself against any of the downsides of AI in the educational field.

Fatima Bey: 

That is really good advice. You have given a lot of good advice today, actually. Do you have any other advice for the youth of the world? AI or not?

Arshavir Blackwell: 

Well, do not be scared of it. It is just a tool and it is a very powerful one. I would say I have always suggested and you know, most students do not learn how to code in school. And I have always thought that that is a very useful skill to have, and I still do, even though AI does most of the coding now for us. Because it makes you think in very logical, formal, mathematical terms. And I am not saying that you have to be a coder, but I think some familiarity with something like Python, which is a coding language, is useful because it gives you a much richer view into what these things are really doing. And you understand why when something fails or when it succeeds or why when you have it do a certain thing a certain way. I get it. It is because the code makes it that way. It is no longer kind of this black box mystery. And I think that that is something that unfortunately our educational system for the most part, I think, has not really done a very good job of helping with because I have not seen a lot of schools that have coding as being even an offering, much less something that is required.

The Importance of Learning AI and Coding

Fatima Bey: 

I think our education system, and this is a very general statement overall, has not caught up with the times and more on more issues than just AI. Antiquated systems, you know, old learning styles that do not work for at least forty percent of the population. I think that that is a problem in general. But that is another conversation. What do you think when it comes to AI is growing? So, you know, when I was younger, they told you get into computer science. What career paths should you be looking at right now when it comes to AI?

Arshavir Blackwell: 

That is a hard question because as we have seen, one thing AI has done, although not as badly as everybody thought it would, has been to harm the computer science side of the economy in terms of coders not being as necessary as they were in the past, because you can get AI to do it. So the problem is it is very, very hard to see where this technology is kind of going to wind its way in the next five or ten years. So I think it is going to actually end up creating a lot of jobs, but I am not sure what those jobs will be. Probably in the near term, I would say going into computer science, it is very interesting. If you are going to go into it for theoretical reasons or because you want to do research, I think that is great. I think unfortunately coding has become much more of an automated process. It is not unfortunate because it has helped us a lot, but it also means people are in that and need to figure out what they are going to do with their lives now that they are not able to get a job as coders. So I think probably, believe it or not, humanities are going to make a real comeback because the ChatGPTs and the Claudes of the world are all systems that interact with people in human language. And we are going to need people that are really good with English and are really good at understanding nuance and are really good at all of those kinds of different things to interact with these new, much more sophisticated systems. So I guess that might actually be, you know, an English major. I mean, who knows? It is not as wild as some predictions I could make.

Fatima Bey: 

Is it because language skills are natural in human interaction?

Arshavir Blackwell: 

I think it goes farther than that. I think that I have read and heard that companies now, because they realize that languages become a more important part of their workflow, that people are not just typing numbers in anymore, but they are actually interacting with these systems. I think having good language art skills and, I mean, I am not saying you have to major in English, but I think having good language skills to interact with these systems is going to become much more important in the future.

Encouraging Youth Entrepreneurship with AI

Fatima Bey: 

That is, I was not expecting you to say the humanities thing. I think that was the mind shift to me. I could kind of see what you are saying. I was going to say, you can let me know if this statement is accurate or not, because you said something similar earlier. Since we do not really know the future of AI, and I think that that is true, because no matter what expert you ask, you get kind of a similar answer. Because the truth is we do not know. The landscape of AI is so rapidly changing. It is like I have never seen anything like it in my life. So what the future looks like is very questionable. But to me, what is not questionable is like you said, learn it. If you learn AI, you have a chance in the future, even if you figure out how to apply it later. But if you know nothing and everything is run by AI, you will still have choices, but they will be limited.

Arshavir Blackwell: 

I mean, I am a big believer in educating yourself as much as possible on any topic, but certainly in any area that has and is going to have so much influence in our day to day lives, I think that is unquestionably correct. And I think almost no matter what profession someone ends up pursuing, they are going to find that AI is a very useful and is probably going to be an integral part of whatever it is they do from day to day. You know, a doctor or a nurse is going to probably be interacting with AI systems many, many times a day. The same for a lawyer. All of those professions are going to be much more AI oriented than they are now. We are just beginning to see that shift. Marketing is another area where AI, I think, is going to become much more prevalent and is already being used quite a bit. So it is hard to think of a profession that is really not going to be affected by AI. I cannot think of any. But it is very much like when the computer, the personal computer, or even before that, just computers in general came out, they pretty much touched every part of our lives, even if we were not necessarily working with them directly. So, you know, what you would do when you went to a bank was so much different after everything became computerized. Then we got personal computers and all of a sudden we were doing our checkbooks ourselves on our computers. AI is going to be very much like that, I think.

Fatima Bey: 

Well, where can people find you?

Arshavir Blackwell: 

A couple places. I have my Substack, as I mentioned, which is insidetheblackbox.ai. And I publish regularly on basically topics of interest about large language models and understanding how they work, which is a big point with me, a big area of interest with me. And also I am coming out with a new application that is a local LLM and that writes in your own voice and that learns how to speak in your voice, your written voice, as well as is aware of all of the things that you have written that you have given to it so you can ask it to compare what you are writing with things you have done in the past. And that is in beta right now. That is called yourvoicecraft.ai. So they are both dot ai addresses.

Fatima Bey: 

And the links will be on the episode page. So you all will see it. And I will say I did take a look at yourvoicecraft.ai. And as a person who writes myself, I found it very interesting. And the fact that you gave literal examples of writing in your voice instead of writing generic. Because a lot of people just settle for generic, which is terrible. But I found it looks pretty good to me. I have not used it yet, but it looks pretty good. And I am going to make a suggestion for our youth. I think that you should take Arshavir's advice, learn AI, in and out, and then start a business with it. There are so many businesses that you can start with AI if you are willing to put in the work. There are. And pay attention to what is current and offer services that people just do not want to be bothered with. You could be eighteen years old and making six figures just doing that if you learn now. Just saying.

Arshavir Blackwell: 

Yeah, I think that is true. And I think that I know actually quite a few people who are solopreneurs, basically individual people who are using AI in order to create their own software stacks and their own companies. So I would say, yeah, right on.

Fatima Bey: 

Well, Arshavir, thank you for coming on. I know there is like a million things we could talk for hours because you really do know so much. But I do thank you for taking the time to come on to my MindShift Power Podcast and I will see you soon.

Arshavir Blackwell: 

Thank you. It was great talking.

Mind Shift Moment: Youth Participation in AI Decisions

Fatima Bey: 

And now for a mind shifting moment. When you take a look at the title of this episode, I really mean it. If you are young right now, twenty four or younger I would say, are you participating in the conversation right now? Are you participating in decisions that are being made about you? Because right now they are making decisions around you, about you, and your children, and you are not a part of the conversation. That is always a problem. So my mind shift moment for you right now, I want you to shift your thinking from being a receiver of decisions that are made and start making yourself part of the decisions. Start inserting yourself in the world of AI because your future and your kids' future depend on it.

You have been listening to MindShift Power Podcast. For complete show notes on this episode and to join our global movement, find us at fatimabey.com. Until next time, always remember, there is power in shifting your thinking.