For one week Charlie Mylie and Jori Sackin had The Big Board at Missouri Bank asking people to make predictions. They greeted the customers as they entered, prompted them to make predictions about the future and then walked them through plotting it on the board. The conversation that follows was sparked by one we had that week. This is the third in a series of 5 that will be published on Ten Millimeters.
J: Would you like to make a future prediction?
P: About what?
J: It could be anything. You could make up your own or you could weigh in on one the predictions on the board. For instance, right now we've got the probability that "the existence of aliens will be disclosed in two years" at a little over 10%. Does that seem right to you?
P: I would put that at 0%.
J: Ok, how about, "The rise of artificial intelligence will happen in two years."
P: Hmmm. I don't know what that means.
J: Which part?
P: What does "rise" mean? Does that mean they're going to take over?
J: Right, so you could imagine "rise" as some Hollywood robot apocalypse or you could imagine it as...well...it's going to get a little more prevalent to the point that most people are aware of it in their everyday lives. Depending on how you define it, the probability may go up or down.
P: So which is it?
J: Let's say it’s the latter. That it becomes more prevalent.
P: Well, it seems like that's already happened right? I mean smart phones and driverless cars and...did you see those google chat bots talking to each other? It seems like artificial intelligence is already here.
J: So I would disagree and the disagreement lies in the word "intelligence". You could think of intelligence as information processing such as a driverless car. It knows when to stop and turn and it knows what speed to go on certain roads. That certainly seems intelligent. I can easily think of a less intelligent car, such as my 1989 GMC Suburban, which doesn't know how to do any of those things, but I think this view is missing something, which is that it’s the success and failure of living in a world that's hostile to its existence and striving to reproduce in that world that ultimately produced intelligence, and so I don't know if you can call something "intelligent" that hasn't been able to pass that test.
For instance if we left a driverless car by itself in the middle of the forrest could we come back in 100 years and find it functioning? No. We also wouldn't find any new driverless cars it had created. Once human interaction ceases with the thing we've built, it fails in a way that intelligence must succeed. The same could be said for the Google bots. Sure, they can communicate with one another. They can form intelligible sentences and sound meaningful, but put them in the woods and see their intelligence put to the test.
P: I don't know. I think you're setting up a pretty impossible scenario. I mean, why do intelligent machines have to be removed from the people who made them? I just...I mean that's not what we're designing them to do. We're designing them to do stuff for us, so of course they're not going to survive on their own. All we want them to do is make our lives faster and easier, so I guess I don't see it as two different intelligences, artificial and natural, that are separated by this huge gap. It seems they're more bonded together than what you're implying.
J: So...yes...you're right. Artificial and natural intelligence work together to create one intelligence that’s made better or worse depending upon their relationship, and so you could imagine a future where this dynamic lifts humanity out of the drudgery of mindless calculation and busy work and focuses our attention squarely on the creative challenges that lie ahead, and you could also imagine a future where it makes us a little stupider, slower, less adept at responding to our environment.
P: But I thought you said artificial intelligence wasn’t intelligent?
J: I think I’m changing my mind...in the sense that it's intelligent when it’s fusing with human intelligence...but...I don't know....I have this idea of something standing on its own, like a parent watching their child grow up. The whole relationship is to get this helpless thing to stop depending on you and to take care of itself, and so when I think about A.I., I guess I superimpose this relationship, though now that I say it out loud it seems silly because part of the fear surrounding A.I. is that it will become unbound from being a tool for human desire, which is exactly what "standing on your own" would mean.
P: I guess I'm a little confused about what the word intelligence means now, because there’s intelligence that’s just information like 2 + 2 = 4 and then there's the way you've been talking about it, which I still don't quite get.
J: Right. So sometimes we're lazy and we use the same word for two different things such as "bank", but in the case of intelligence, I think the two ways we use it are deeply intertwined. So we have intelligence as good information, like 2 + 2 = 4, and we have intelligence as successfully being in the world. To put it another way, we have intelligent information and intelligent action. This is not so unusual for a double meaning of a word to be both the object and the action of that thing. Think about the word "drink" or "smoke" or "fish". They all can be the object as well as the action associated with that object.
P: This is starting to get a little abstract for me. Give me a concrete example.
J: Sure. So, what did you eat for lunch today?
J: How did you decide on tacos?
P: I wanted to walk somewhere because it was nice out and I originally thought I wanted Vietnamese...but once I got there I remembered there was this Mexican place down the street, so I changed my mind.
J: Great. So that sounds like a pretty simple story, but let's look at it with the tools we have so far. First, we have intelligent information, meaning you had good knowledge of the possible places to eat in walking distance of where you work. You have this information because you've had to solve the "lunch problem" before and have experienced and remembered the consequences of your decisions. A lot of that information you've packed away and so I imagine it feels fairly intuitive to access. You didn't for instance consciously think through each individual lunch option and weigh the pro's and cons of each place.
P: No. That would be exhausting.
J: Right, so I imagine you skipped over quite a few places that have ceased to even be considered as possibilities such as the Subway on the corner.
P: I love Subway.
J: Ha. I guess I made an assumption about you...but...ok...I'm sure there are places you've eaten and then decided to never go back again or places that looked so unappetizing you would never think about them as options.
P: Sure. That's true.
J: I mean you already eliminated a huge swath of options by deciding you wanted to walk. That decision took almost 95% of your lunch possibilities off the table. Ok. So you walk down to get Vietnamese but then you end up eating tacos. That's quite a shift. Can you remember what happened that made you change your mind?
P: It wasn't so much a comparison of options. At least I wasn't conscious that was what I was doing. If I think back, I had a moment where I imagined eating a taco and really enjoying it and it was just this simple strong visual thought. That's all it took.
J: So you imagined yourself eating a taco and that gave you a certain amount of pleasure and suddenly the Vietnamese didn't seem so good?
P: I don't know. I just didn't think about Vietnamese anymore. Actually, that's not true. I remember having the thought that they make you pay at the register, and for whatever reason it feels awkward standing there with the receipt waiting to pay. I don't know why, but that thought popped in my head, so I picked tacos.
J: Ok. So you compared two unrelated variables, um, waiting to pay for your check vs. the pleasure of biting into a fresh taco, and then you went with the one in the moment that made you feel good?
P: Right, but the thoughts were visual experiences. I imagined myself there. It wasn't visceral, but it was definitely bodily.
J: Sure. And did you enjoy the taco, like you imagined?
P: No. Not really.
J: Do you think you would go back again?
P: I mean it wasn't terrible. It was just a little expensive and I remember feeling bad for paying more than I usually do for lunch and also not really enjoying it. That felt.....I don't know how to describe it. It was subtle but it was definitely unpleasant. So I guess I wouldn't go back now that I think about it.
J: So the next time it's nice out and you decide to walk to lunch and you mentally scan your possibilities, would you consider the taco place an option?
P: Probably not. At least not for awhile. It's funny now that I think about it, I've had this experience before, of eating there, being disappointed, telling myself I wouldn't go back, and then a certain amount of time passes and I think, "Maybe I should try that place again", and then I try it and the whole cycle starts over. Not very intelligent I guess.
J: Well, so you didn't have an awful experience. The waitress didn't trip and drop tacos on your head. You didn't get food poisoning and were throwing up for hours. It was a slight irritation and so maybe given your particular cognitive capacity, it just wasn't worth remembering.
P: Or I could've just forgot.
J: Sure. You may have a bad memory for certain things. I have a terrible memory for people's name. How much do you really care about food?
P: What do you mean?
J: I like eating food, but I tend not to think about it that much. I don't cook. I don't think about food preparation or try to dissect the ingredients when I eat something, so I probably don't have a good memory surrounding that area because I don't put much thought into it. Don't get me wrong. I love a good meal. But I often forget to eat and solve the "lunch problem" as quickly as possible so I can get back to doing the thing I really want to do.
P: Yeah. Ok. I enjoy food but I don't really think about it that much either.
J: So the point is that you had a particular problem to solve, "eating lunch" and you navigated around successfully to accomplish that task.
P: What do you mean successfully?
J: So you wanted to accomplish a goal, "having lunch", and you accomplished that goal. On your way you didn't step into oncoming traffic, you didn't forget your wallet and you didn't walk into the restaurant completely naked. You accomplished what you set out to do and in the process you gained intelligent information, which is, the phenomenological experience you described of eating tacos and being disappointed. That information is given a value and then either stored or possibly discarded as "unimportant". Given the conversation we're having now I doubt you will forget this experience again. The value of that information has risen, or at least, your attention has fixated on it for long enough so as to engrain it deeper in your memory. So this is the intimate relationship that I was talking about earlier between the two definitions of the word intelligence, which is, that intelligent action produces intelligent information. The more you act intelligently in the world, the more intelligent information should be generated, and consequently, the more information you will have to sort through to find what is relevant to the problem at hand.
P: So you're saying that forgetting my previous taco experiences was unintelligent right? I made a decision that wasn't particularly satisfying and then I made it again and again, and for whatever reason, I didn't update. I mean, I feel like I have a pretty good memory, but I guess I don't for some things.
J: Well it probably depends on what we're talking about. I have a terrible memory for how things look because I have a particular type of cognition, which is, I have very little internal visual information that comes to me when I think. So when I recall a memory such as eating lunch there's usually not any visual information present. I can tell you a story about it. I can recall certain details, but those details aren't remembered by internally perceiving a picture in my mind. Consequently I'm bad at remembering visual details about things, such as what people were wearing, paintings that were hanging on the walls, things like that.
P: Weird. I've never heard of that. When I think about eating that taco I actually see myself doing it. I see it right now. I can see what I was wearing and what the table looked like. I can even see the people at the table next to me.
J: Right. So all of that information is unavailable to me. So that's quite an impairment given that particular task. However, there are things I can do to supplement my low ability for visual memory. I can write stuff down. I can take pictures with my camera. That's a kind of artificial intelligence because it's an add on to my conscious experience that modifies how I think and what I'm capable of remembering. I know I have a low capacity for memory, so instead of trying to strengthen that weak ability, I develop a strategy so that information can be stored in a way that doesn't require the cognitive effort I know I'm not particularly suited for.
P: That's interesting. I've never thought about writing and taking pictures as a supplement like that.
J: Yeah. I've also noticed there's quite a difference between thinking silently in my mind and talking out loud to myself. When I think silently it tends to loop in ways I don't particularly find useful. There's often lots of repetition and my train of thought gets lost easily. But when I speak those same thoughts out loud, a lot of that disappears. I'm more confined to the nature of what it means to speak in sentences, that they need to be finished in a coherent way, even if no one is listening and so I tend to stay on track. This is a cognitive trick I learned that circumvents my limited attentional ability but it does take extra work so I don't pull it out for any old thought. I really have to value what I'm thinking about in order to justify the extra effort.
I could say the same thing about writing. Writing is a context where thoughts materialize in words and then stay put, and so it's much easier to build ideas in this context because I no longer have to go through the cognitive effort of having to remember all of that information, to keep track of where I am in an argument or what comes next. It's an incredible invention that dramatically changes the way I think, act and respond to the world, but it seems so mundane because most everyone in the world can read and write now, so we tend not to think about it too much.
P: So let me get this straight, you're saying that I didn't remember my bad lunch experience because I didn't value the information enough, right? I didn't take the effort to really engrain that in my mind as something worth remembering and so it probably stayed around for a day or so and then eventually wasn't active in my mind. The failure of making the same mistake over and over again wasn't painful enough to really stick with me and so I just didn't put the effort in. But that just makes me think about all of the information I take in on a daily basis that I don't really find important. I mean, as I was walking to lunch I passed by this yellow Hyundai and I remember thinking, "That's a stupid color for a car." Why the hell did I remember that? That's seems completely inconsequential. Maybe my filter is messed up.
J: So when you act intelligently in the world you gain information, but like you said, information is coming at you all the time. It never stops. So you used the metaphor of "a filter" which I think is a good one, because how do you decide what's important and what's inconsequential? Well, it depends on the problem set you're working with. When we go around trying to solve the lunch problem there is a huge amount of information that suddenly becomes inconsequential, such as the color of my shirt, how many leaves are on the magnolia tree by my office or that I have curly brown hair. We intuitively filter out that information and effectively deal with what we believe to be relevant, which is, what restaurants exist in walking distance, how much do I enjoy them, how much do they cost, how much money do I have, etc. Effective filters can intuitively identify the relevant details that are involved in making an intelligent decision. If, for instance, I failed to think about how much money I have in the bank and then go to the most expensive restaurant in Kansas City, that would be an unintelligent decision
P: I think I've made another unintelligent decision, which is, I've stayed too long at the bank having this conversation. I'm actually late for work now.
J: Sure. I really think we just skimmed the surface, but if you want come back tomorrow. I'll be here till Friday.
P: Yeah. It was good talking to you. Let me think about all of this and I'll stop back by tomorrow.
J: Great! Thanks so much for taking the time to talk!
P: Sure. I'll see you tomorrow.