Vignette on Conceptual Intelligence Effort

      No Comments on Vignette on Conceptual Intelligence Effort
ShareShare on FacebookShare on Google+Tweet about this on TwitterShare on LinkedInShare on RedditEmail this to someone

This post is a planned excerpt book I am working on that addresses the topic of machine intelligence. This section is written as a fictional vignette about a small team working on a breakthrough in the area of intelligent systems. This content is a very rough draft. I would love to hear any feedback…thanks…

“I’ll be right there!” Jennifer sprang from her desk chair and headed towards the front lobby of the office. She had a habit of checking her hair before greeting important guests but her excitement overrode the protocol this time. In the lobby, she greeted Michael, the private investor funding her project, with a handshake and a big smile. Trusted colleagues had informed Michael of a small and talented team that was doing some incredibly interesting work in the area of artificial intelligence and machine learning. Several meetings and slide decks later, Michael had decided to financially back their effort.

“I hope you are really floored by what we show you today”, Jennifer said as she led him the to test room. The door to the test room was marked with an eye-level sign that read: “The Cave”. As they entered, the three other team members stood to greet them.

Jennifer: “Walter, Cindy, and Todd, we want to thank Michael for being able to see our demonstration on such short notice. Michael, I believe you remember the team behind our work. I can assure you that they are really looking forward to showing you where we are today.”

Michael: “Of course, hello, everyone. I’m really looking forward to this demo as well. Jennifer hasn’t said much but I know her well enough to realize that this is something momentous in her assessment. Let’s see what you have to show me!”

Jennifer: “Great! We’ll get right to things. Let me explain everything you see in this room. Ok, on this table is our prototype system. We’ve nicknamed it “Billy”. Billy is basically a computer, a high-end enterprise-grade machine, hooked up to this 3D camera that feeds him sensory images from the environment. This laser pointer is also connected to the computer. The laser pointer is mounted on two servomotors giving it two degrees of freedom. Basically Billy can position the laser pointer anywhere on the white wall that the camera points at. The refined system output will be displayed on the monitors to its right.”

Michael: “A standard computer with peripherals: the 3D camera, the laser pointer assembly, and monitors.”

Jennifer: “Yes, and most importantly, the computer runs the program that constitutes the mind. It allows Billy to create, retrieve, and apply concepts. That’s the crux of the project.”

Michael: “That’s the part I’m having trouble understanding well, to be honest, despite your great previous explanations and videos with the colorful graphics…”

Jennifer: (Laughs) “Well hopefully today’s show will help with that understanding. Let me explain the other equipment in this room. This separate computer and projector to the left of Billy is going to deliver the universe to him. Basically we are going to display simple shapes on the wall opposite the camera and do some simple manipulations with those shapes. Now those two large monitors to the right of Billy are going to help us understand what is going on in his mind. They will display numeric and graphical representations of the concepts going on in his mind. We’ll explain that better as the experiment proceeds.”

Walter: “Ok, we’re wiped clean and ready to go.”

Jennifer: “Great, so we’ve just given Billy the ultimate mind-eraser. It’s happy birthday for Billy — his mind is completely empty. He is just running the code that allows him to learn, but right now, he has no knowledge of the world or anything.”

Michael: “So he is wired to learn but doesn’t have any…concepts…”.

Jennifer: “Exactly. So the camera has engaged and he is staring at that blank wall. You can see on the monitors….those long strings of numbers slowly scrolling…that’s Billy thinking.”

Michael: “What do the numbers mean?”

Jennifer: “They don’t mean a whole lot now. You can see on the monitors that we aren’t finding any real patterns. At this point, Billy is scanning everything he can see and coming up with white nothingness.”

Michael: “It reminds me of that scene in ‘The Matrix’ where the green numbers are scrolling…”

Cindy: (Laughs) “We have said the same thing. Actually, the comment isn’t far off the mark. These long strings of characters and numbers really represent a more complicated, composite mathematical structure. If this number string mirrored the activity of your brain it would indicate a specific set of neurons that are currently firing and how those neurons are sending signals to other specific neurons.”

Todd: “I will display the square now.”

Jennifer: “Good. So now we are projecting a small black square on the wall. Look how the numbers change are we are definitely seeing a pattern repeated.”

Michael: “That pattern represents the concept of a square?”

Jennifer: “It really represents the concept of a ‘thing’. Billy doesn’t know enough about other ‘things’ right now to really understand a square. But we’ve replaced the nothingness with a bit of information amidst that empty space. On the monitors you can see the word ‘Thing’ displayed. Based on the testing we have done so far, we believe that at this stage of the process, the repeatable pattern represents the concept of an item that is bound, in this case, in two dimensional base. Now we’ll display a small black circle.”

Michael: “I see some new colors and patterns on the monitors.”

Todd: “But the patterns for ‘thing’ still appear for the circle. He knows it’s a different kind of thing but he’s also starting to generalize.”

Jennifer: “Now we’re going to display a triangle. You see some of the same patterns appearing on the monitor. But this time we’re going to change the size of the triangle: it’s going to expand and collapse slowly several times.”

Michael: “Is the system recognizing ‘change’?”

Jennifer: “Yes, we’re pretty certain that at this point the patterns represent change. Change is a very fundamental concept, even for us. You’ll soon see how it supports many more complex concepts. But first we’re going to try to introduce another important concept. The triangle is now disappearing from the display and reappearing in different locations on the wall. So for the first time, we are emphasizing position or location…”

Michael: “So you just introduced the concept of change and the concept of position…I may be able to guess what you’re going to do next?”

Jennifer: “Hah, probably, very good. You see that the square now appears on the far left of the display and we are slowly moving it to the right…”

Michael: “And the monitors are showing that the concepts of thing, position, and change are being recognized now.”

Jennifer: “The combination of these concepts comprises the slightly more complex concept of movement. So now Billy has a conceptual representation of movement that can be retrieved and applied in other contexts.”

Michael: “ I don’t know if this is all smoke-and-mirrors but it’s starting to get very interesting.”

Todd: “We aren’t far from the point where we can allow outside teams to create their own test cases and we can help them analyze the output data to show how concept patterns are corresponding to the input they deliver. So we will be able to prove the capability of the program.”

Michael: “Then you all will start looking for yachts on eBay…”
(The team laughs)

Jennifer: “So now the big finale for today. We’re going to cheat a little bit. Todd is going to trigger the system so that Billy becomes aware of his laser pointer appendage. Keep in mind he won’t know how to use it at first.”

Michael: “Ok, the laser pointer is wandering around. It’s flailing a bit. It kind of reminds me of a baby trying to move its arms…or maybe it reminds me on an inebriated person. What concepts are we seeing on the monitors?”

Jennifer: “A lot of the established ones: thing, movement. The red dot from the laser pointer is a new ‘thing’. Now something important is happening — Billy is realizing that he is moving the laser pointer. He is correlating his motor actions with the position of the laser pointer. Effectively he is learning the concept of cause-and-effect.”

Walter: “You can see that the movement of the laser pointer is becoming much more fluid.”

Michael: “Yes, I noticed. He got control of his mechanics more quickly than the baby I was picturing. Why is he moving the laser pointer around? What is the motivation?”

Jennifer: “Ok, perfect lead in… Right now he’s just moving it around trying to learn about his world. Although we don’t preset the program with concepts, it is designed to simulate the mechanics of the brain. It is programmed to pursue the collection of environmental signals — perceptions — and to make sense of them.”

Michael: “So it’s inquisitive.”

Jennifer: “Essentially. In order for the system to be truly intelligent it has be able to incorporate, really learn, new concepts based on its environmental perceptions in conjunction with the concepts it already knows. This is the same we humans learn, whether those environmental perceptions are things we’re taught or things we observe.”

Cindy: “Now we have to wait for it to happen…”

Jennifer: “Yes. So, what are we waiting for? Intelligent beings are goal-driven. Some of those goals are given to us as assignments and some of them are more naturally instilled. When we are hungry we have a goal to eat and when we are tired we have a goal to rest. We gave Billy a built-in goal concept. When he achieves his goal, we will send him an input signal indicating so. But he doesn’t yet know what he has to do to achieve that goal.”

Michael: “What does he have to do? You can whisper if you don’t want him to hear me.”

Jennifer: “Hah, we don’t handle audio sensory input yet. He has to hit the shape on the wall with his laser pointer. This usually takes a while…ok, he did it. And that makes the shape disappear.”

Michael: “Ok, I see the “goal” indication on the monitors. The next shape is displayed but he isn’t putting the laser pointer on it.”

Jennifer: “He hasn’t realized that it was that action that caused it, although he remembers that he did perform that action. Note that this is a different shape in a different position on the wall. He is retrying his actions now. He can conceptually recall those actions by position, direction, and so on. Ok, he got that shape. Next one displayed.”

Michael: “I’m seeing all the concepts now on the monitor: thing, cause and effect, reward.”

Cindy: “And the color-coding on the monitor display help us see that he is bringing the laser pointer ‘thing’ and the shape ‘things’ specifically into his concept context.”

Michael: “He’s performing the action more quickly now. So he gets it?”

Jennifer: “Yes, he gets it already. Eventually we are going to make things tougher for him. He will have to pulse the laser and do other variations based on the type and color of the shape. But this accomplishment is a golden spike for us. The system formed, retrieved and applied concepts on its own in order to achieve a goal. We know of nowhere in the world where this level of autonomy has been achieved.”

Michael and Jennifer left the group with congratulations and encouragement. Michael offered to take Jennifer to lunch to discuss what he witnessed and what he could expect in more detail. Jennifer warned him that he’d better be prepared for a long lunch.

Michael: “Back at your office, I was wondering if I was seeing history in the making. How did you guys accomplish something that has been such an elusive goal.”

Jennifer: “We knew from the start that this wasn’t a software-only problem and we’ve very deliberately involved experts from several fields. Cindy has expertise in neuroscience and Todd has extensive education in pure and applied mathematics. Walter is an expert in machine learning and complex algorithms. All of them are competent software engineers as well. They work well together and have also brought in neuroscientists and mathematicians as needed for discussions and specific work.”

Michael: “What mathematics disciplines are involved to support the program?”

Jennifer: “We’ve really developed a new type of mathematics that we’re calling Concept Mathematics. Instead of equations with operators and operands, etc. we overlay these mathematical concepts against each other and the result is another concept. As we mentioned to you, those long character strings that represent concepts are complex mathematical structure. The structure contains numeric “edges” that associate the concept structures to each other, similarly to how synapses connect neurons.”

Michael: “This new mathematics is modeled on our understanding of the human brain?”

Jennifer: “Yes, it is. There has been skepticism around a computer’s ability to achieve human brain-level intelligence. A lot of the doubt relates to the amount of power a computer requires compared to much more power-efficient human brain, as well as the massive parallelism the brain performs in processing information. We attacked the problem from a different direction: instead of addressing these challenges with a new computer architecture, we created this new mathematics to reduce the work the computer processor has to do.”

Michael: “Well I’m really impressed with what I saw. What’s next?”

Jennifer: “We’re going to introduce Billy to more complicated concepts. We’re going to bring in a data scientist who can help us compare the environmental stimuli we deliver with the conceptual outputs so we can better understand how the system models and leverages concepts numerically.”

Michael: “You don’t fully understand what the system you built is doing…?”

Jennifer: (smiles) “We’re in a new world of autonomy…this is the downside. We’re going to rely on observation and measurement to really understand the system’s behavior. We are also anxious to start processing audio sensory input and managing concepts at the level of spoken conversation. Can you imagine having a conversation with Siri or Alexa at the conceptual level, and letting that device or service search and conceptually manipulate an internet of information to deliver answers to you?”

Michael: “That’s the new world everyone’s talking about.”

Jennifer said nothing and waited for the question she wanted to hear.

Michael: “Ok, tell me what you need.”

Leave a Reply

Your email address will not be published.