Artificial intelligence is all around us. We use it for everything from researching topics we’re interested in to asking it for help when we’re writing something. In this program, we learn some of the basics of AI. What is it? What is its history? Why is it being developed? How is it evolving?
This is the first of a series of conversations about AI that we’ll present this summer and fall. We’ll talk about AI’s potential and its dangers? How should we best use it? And what should we be aware of when we do?
We’re joined by Grant Erickson, the CEO of IntelliTect, and Graham Morehead from Pangeon and Gonzaga University.
This transcript has been edited for length and clarity.
Doug Nadvornick: As two guys who work in the AI field, how quickly is this moving?
Grant Erickson: One of the guys that we have on staff, who tracks this probably more than anybody else, uses the comment that says, AI is doubling in capability every seven months. That's approximately the level that we're at. Graham, s that about the same numbers that you're seeing?
Graham Morehead: The rate is getting faster and by the way, there's a lot of jokes right now. People will say things like, wait, you're still doing that? Such and such has been the industry standard since 2 p.m. yesterday. It is crazy. When I was just starting out in this field in the nineties, you would have something huge thing come along every 10 years, it felt like. Then every five, then every two, then every year. And now every day I wake up thinking today there might be a brand new amazing invention or discovery in the field of AI. It is sometimes two or three a day. It is unbelievable.
GE: It really is shocking because I teach it at Eastern, a class in web development, and I taught this last winter quarter. I started prepping over Christmas because I've taught it several times. I started saying well, can I just ask AI to do the final project and it did it. And so I spent a week over Christmas rebuilding the entire class around what it would be to use AI.
Really, I think I would put for us in our business, November 25th, 2025, was the day when things shifted. Quad Opus 4.5 was released and that's really the model that made it so that as software developers, we could be significantly more productive with AI than not. Then the next one came out, 4.6, in, like, February and 4.7 just came out about a month ago. So we're seeing significant increases in what's possible, although we're kind of wondering, hey, are we going to hit some kind of plateau? Because now we're finding that, hey, they're not getting much better for the basic tasks, but we're starting to be able to ask them to do more complicated things and they're able to go longer without as much human intervention.
DN: What is artificial intelligence?
GM: Intelligence is, can you take some inputs and give the right output? With text or language models, typically you have some question you can formulate as text or code and it will give you some output, which is also text or code. That's intelligence.
DN: Going back to artificial intelligence in the 1990s, how was it conceived differently or perceived differently than it is now?
GM: It kind of started with Alan Turing in the '40s. Some of the papers that he and (Alonzo) Church wrote in the '40s are still important. We still read them today. The Turing test was a test for, is something actually intelligent? He envisioned a world where you could text to a human and a computer and not be able to tell who is who and we're in that world now. When I first started teaching, we weren't in that world yet. There were things I would say in the course, this is something AI can do. This is something AI cannot do. Only humans can do it. That box is getting smaller and smaller and smaller on that side.
But I would say there's many different kinds of intelligence. There's linguistic intelligence, there's math intelligence, there's artistic (intelligence), there's physical intelligence. All of these are things that AI is expanding into. There's a lot of differences between them, like, what is good art? Art should be, I think, a relationship between two thinking humans, the person seeing the art and the person making the art. AI really should just be a tool. But AI is trying to expand into all these things.
DN: And why is it important? Why is that important?
GE: I think the way that I've thought about what AI does is AI is able to look at something, like the capital of France, and then it's able to predict the next most logical word in this case that's going to come out. We see that happening across whatever it is, maybe it's music, maybe it's generating digital images, generating words. That's really what AI does. It's a logical predictor based on all the training data, which is essentially everything that folks can find to train it on the entire internet, plus bunches of things that allow it to be very good at predicting the next thing and it's able to hold way more things in its "memory" than a human could. I's able to pull lots of data in and make lots of inferences and the reason that's important is it allows us to do things and come up with things that we as humans just have a hard time doing.
We often will specialize in a particular area. We may have a real expertise in sports, but we don't have this deep expertise over in some other area, being able to take those things and combine them and then be able to generate things like ideas. For example, I have a friend that does puppies and she wanted names for all the puppies and it gave her names with all these different fun things to be able to come up with that. Now that that's a fun kind of non-business use case, but I think you can pretty easily pivot that into a business case and think, how do we optimize things that we just haven't been able to optimize in the past?
I oftentimes use the example of pickleball. Pickleball, relatively new sport invented here in the state of Washington, it's kind of taking the world by storm. The last three years, pickleball moved from a side sport to a really competitive sport and I think what we're seeing and the reason this is important is AI is allowing businesses and allowing people to optimize things in ways that were not possible in the past.
GM: Steve Jobs would say the computer was the bicycle for the mind, AI is a motorcycle.
DN: So when did we start thinking that AI might be a bad thing?
GM: 1968 when “2001: A Space Odyssey” came out. People have different perceptions of the relative good and the relative bad of AI.
DN: What to you is the worst case scenario?
GM: The worst realistic case scenario is that AI doesn't harm anybody, but it just becomes something we lean on too much and then we all become stupid.
DN: Do you think we're headed that way?
GM: There are people who will head that way. Yes. We don't all have to choose that.
GE: I was talking to several groups of students and I use the "easy button" example, the big red "easy button." I said, you have the biggest temptation that has ever been in the world of education because you can literally reach out and press that "easy button" to write that paper and then go out and play with your friends. The thing is writing is hard. Doing math is hard. All of these things are hard. The purpose is o learn how to think and so, by pressing that "easy button," even though it gets done faster, you're robbing yourself of the chance to build the neural pathways, to think logically through things. I never realized that we don't learn math and algebra and calculus so that we can do those things. We learn them so that we learn how to think and I think if we lose that, we're in trouble.
Now to double down n what Graham was saying, if you're just using Chat GPT and you chat in, Hey, what about this? I mean, what's the worst it can do? It can like give you some kind of really horrible response, right? But what we're talking about now is, what if I want to give it access to my email? Maybe just to read to start with. What if it could just like respond to some of my emails? What if I needed to do some shopping for some lighting fixtures (for my house)? What if I gave it access to my credit card and it can go out to these lighting stores and do this stuff? Wouldn't it be better at managing the power grid than us? It could choose targets for the military way better than humans could. All of a sudden we've given AI agency to do things and I think this is where, without some controls, things can start going off the rails because AI is what we call non-deterministic. You don't know exactly what you're going to get in for the same inputs. You may get different outputs.
GM: That's why I want to bring back the old fashioned AI because old fashioned AI is deterministic. There's always two halves of it. This argument goes back to the 1950s when AI first got his name ,1956 at a conference in Dartmouth, New Hampshire. There were guys on both sides of this. Marvin Minsky was one of the leaders of the logic side, what we now call type two AI. Francois Chollet is a current researcher who tries to come up with what are the questions that are so hard that AI can't answer them, but humans find them easy. He calls these arc AGI tests and he coined the term, type one and type two. Type two is the logic we just said. Type one is when you just take millions of little things, put them together, connect them up and train them and who cares how it solves the problem? It just, it solves it. Neural networks are an example of type one, what we used to call connectionism. The first guy who did this was Frank Rosenblatt in 1958 and he didn't use a computer. He used wires, hundreds of physical wires and electrical connections between them. So you have the connectionism, which in a sense is inspired by the brain, millions of things connected up., and then you have the logic. The scary thing about the type one is that even when it gets the right answer, it's probably for the wrong reason.
DN: If people come to you and ask you what is the best way for me to use AI, what do you tell them?
GE: I wish I had a nickel for every time somebody asked me that.
I think really it depends on the person. For example, my wife was just excited that she could figure out how to make our lawn greener with AI. She literally went out and took some pictures of our lawn and fed it into AI and she was able to really get a handle on, Hey, you have this, you should be doing these things. And hey, our lawn looks way better than it used to. I think that's some pretty easy, low hanging fruit for the most part. I think we're going to see that evolve relatively quickly.
Again, it's this whole agency thing. How much of my life or my digital life am I willing to let AI have? It's hands on or it's digits on. It's going to vary depending on what a person is looking to get from it. What we're seeing now is what I call the wild west. Everything's everywhere and everybody's looking for the gold and all that kind of stuff. We're just seeing the tools come along that are packaging AI in such a way that it's much more consumable because everybody kind of knew about the chat things. But that's just the surface. It's actually a pretty horrible interface when you think about it, having to type stuff in and read it back. But if you can get some intent and things like that, it can actually start doing some pretty useful things.
DN: We're right in the middle of steering how we want to, as a society, deal with AI. How would you counsel elected bodies who are thinking, what are the guardrails we should put on this? What are the things they should be thinking about?
GM: There is no one answer, but first of all, we've got to recognize how people feel about all this. Eric Schmidt just spoke in Arizona, you know, at a graduation ceremony. He got booed every time he said AI. Now, my daughter is 20 and her generation, they don't like AI encroaching on artistic things at all and there's a whole lot of people who just, at that age, don't want AI and we need to recognize that and understand why.
But at the same time, this is a tool that we don't want to just throw away. It's not going to be thrown away. The world is going to use it and we need to be able to leverage it. It's a lot like electricity. When electricity first came into this country and the internal combustion engine at the same time, it created a lot of disruption in the economy. A lot of jobs were gone. People were killed by electricity. People were killed by motor vehicles. But we still recognize they are a net benefit and I choose to live in a house that has electricity rather than out in the woods and most people do make that choice every day. We use electricity even while we're sleeping. The same thing is going to happen to AI and i's going to be little by little and we're all going to depend on it. We need to understand it.
But the electrical world is not without regulation. There's a lot of regulation and there's a lot of people who are certified to do X, Y and Z in the world of electricity. There may be something like that for AI at some point. I hesitate to say we should slow anything down, though, because anytime that you force too much restriction in one thing, you tend to make just a few big companies powerful and all the little guys have no chance.
DN: So having paid attention fairly recently to the discourse about it, tech guys like you tend to be bullish about it and regular people who don't deal with it tend to be bearish about it. How do we find that middle ground so that everybody feels like we're pushing it the right way?
GE: Yeah, it's really interesting, people's responses to AI, like how do they feel about it? Even inside our company, we have different responses along kind of a spectrum and I put on one side, we have the people that write software because it's a craft and they love the solving of the algorithm and how do you optimize this and make it fast and beautiful and elegant? And then you have the other people that just love getting stuff done and building something that changes people's lives and makes them better. And hey, you can now care for more people because you're using this software. Depending on what side they're on, you either just took away the thing that they hold most dear and they find as part of their identity, this building of beautiful software, it's a death. It is so hard for that transition. And then the other people on the other side, wow, I can just do more stuff. I think you have these two responses.Is this useful? But then what is it going to do to me and how is it going to affect what I'm doing and the things that I really care about and the things that I find a lot of joy in as part of being a human being.
There are very few limits on how fast this is moving, how fast we can build data centers, essentially. W've seen it go incredibly quickly and I think we, as a collective society, will be studied in the future to see how we did, how did we do this? Because when you think about even something like social media, it took us at least 10 years to figure out what some of the impacts of social media are. This has an adoption level of social media in like two years. We have a higher penetration of the number of people because basically everybody, if you're using Google, you're using AI. So everybody's using this and we don't know and really understand what the long term ramifications are yet.
DN: I know a lot of people worry that people are going to become obsolete because the intelligence and the computers are going to take over. Is that a myth? Is that an unreasonable scare tactic?
GE: I think it's something that we've seen throughout history, right? I think about a situation like a Downton Abbey. I enjoyed that show and I loved how there was a place for everyone. Everyone in that little microcosm had a thing that they could come and contribute. I mean, going through kind of that being the tail end of the industrial revolution. I think what we're seeing is that we're pushing humans to do more and more and more and so what's happening is the act of writing code. For us, that's one thing, but now AI is able to do that part. So I'm pushed up a level and I'm now contributing at a more creative guidance. What do I really want? And what's the real value as opposed to going down and doing those kind of more repetitive type tasks. I think that's what we're going to see. The issue was going to be the people who find themselves able to step up to that next higher level are going to benefit immensely from AI and the people that ,for whatever reason, either decide not to or can't, it's going to be significantly harder because I think it's going to be harder to find a place to provide value into society. I think those are some of the tricky questions that we as a society are going to have to answer.
GM: But I think they will go up the value chain, just like you said, just like we've been doing for thousands of years. As you get inventions at some point, you don't have to be the one to pull the plow anymore. We have a large animal for that and we keep moving up the value chain. Think about farmers of today, they have a GPS-controlled giant combine that has either little lasers or electrical fields that will shock the weeds as you go along. There's a whole lot of things that enable us to have only one farmer per like thousand people in this country, which is incredible. We used to be mostly farmers just to barely survive.
There's a scene I like to show my students in a movie, The Devil Wears Prada, where they talk about this sweater, the cerulean sweater, and this character is wearing a sweater that's a certain kind of blue and this other character says that that is because of this other thing that came before and this other thing that came before it and this other thing that was decided upon by people in this room. They are the tastemakers. They decided what is good and what is bad for the next season and that rippled through society. That is something that is a job for a human. What is good? What is bad? What should we make? And then maybe the AI can do a lot of the making, but the human is the one who is the tastemaker. We all need to become tastemakers in one of the infinity number of fields that we all contribute to.
DN: So let's go back to the question of regulation. There are people in Congress who want to regulate this. You have a president who is not real eager to regulate it. Alot of folks in the states say, fine, Congress, if you're not going to regulate it, we're going to do it at the state level. So what's the best way to do this? Or is regulation just not feasible. We should just let it go and let the market decide how AI moves.
GE: I think Graham made an interesting point is that regulation, in a situation like this, we're so early. If we try to regulate, we will do it wrong at a certain level and push it into the people that are willing to ignore the regulations or willing to kind of work around them. So you're going to end up with a few people that have all of the toys, which I think is probably in this case, a worse scenario than having a much more competitive marketplace, because honestly, a new company can pop up just all over the place. AI is now building on itself and there's other countries and people that are after other things that are also building AI. Now, how we want to play all that. I would never claim to understand how all that works. But to me, it feels like we need to be talking about these things.
We use a three-piece framework. We need to have a win for our clients. We need to have a win for our company and we need to have a win for our employees and if any of those people are left out, we have the wrong answer.
GM: The point of regulation is to punish things that cause great harm to lots of people and encourage things that cause great good, and it's kind of an art to find out where that is.
One way AI has affected us all is it took the data that we gave it in our written words, in our pictures, and now these profitable companies are benefiting and we didn't get the benefit in that at all. I'm not a proponent for UBI (universal basic income), but I am a proponent for us getting some kind of dividend from the great value that we all created together as a human species. ome of the training data goes back hundreds of years. So regulation could come in there somewhere.
