Ask A Genius 656: Rick’s Thoughts on Artificial Intelligence
Author(s): Scott Douglas Jacobsen and Rick Rosner
Publication (Outlet/Website): Ask A Genius
Publication Date (yyyy/mm/dd): 2021/12/29
[Recording Start]
Rick Rosner: Before we start, I want to clarify that my understanding of AI and machine learning is quite basic, as I’ve never taken a formal course in these subjects. However, as we discussed last time, eight years ago AI wasn’t a common term in most people’s vocabulary, though it has now become a significant part of the tech marketing sphere. Products are often marketed as having AI, which seems to be more of a selling point than an accurate description. ‘Machine learning’ might be a more precise term for what these AI products actually do.
When we try to define AI, it might be more useful to consider machine learning, which essentially involves circuits with feedback mechanisms. This means that circuitry or decision trees leading to better outcomes can be tuned to receive more traffic, right? The idea is that successful circuitry is reinforced while unsuccessful circuitry is suppressed.
The core technology behind a machine’s ability to improve its performance involves running simulations repeatedly so that the circuits can be refined.
To lay the groundwork for our discussion, let’s recall our previous conversations about the brain and neuroscience. The brain acts as a predictor and a preparer. Its job is to anticipate what’s going to happen and position the organism it belongs to for optimal response. The brain generates various possible outcomes and aims to prepare you in the best way possible. There are different strategies it might employ, like a high-risk, high-reward approach, where, based on the brain’s model, there might be a 20 percent chance of a significant payoff. A more conservative brain, however, might steer you away from risky situations in favor of safer alternatives.
All these predictions and decisions are grounded in the brain’s understanding of you and the world. Essentially, we are evolved to survive and improve our circumstances. This includes basic survival instincts and, in some cases, altruistic behaviors towards kin. The brain positions us accordingly, like triggering adrenaline for fight or flight responses.
Because the brain has what, about 10 to the 10th neurons, I think? And each neuron, on average, has around 10,000 dendrites, so that’s fairly complex. It’s not impossibly complex, though, especially when you consider that we’re starting to see circuits with a number of bits that are roughly equivalent to our neuron count. We don’t fully understand the information capacity of neurons to make exact comparisons, but at these magnitudes, it seems feasible that we could manufacture devices with similar complexities. I’m not sure if the focus is currently on creating conscious machinery, but if it were, and we knew how, we probably have the technological means to do it, in terms of assembling enough circuits.
From what we’ve learned through machine learning, seeing how neurons and circuits feed back on each other, it doesn’t seem technologically unfeasible to eventually, maybe even within decades, build machines that think in ways similar to us.
It would certainly be beneficial to have a mathematical model of consciousness and the mind. Even without such a theory, it’s possible to build machines that exhibit consciousness through reasonable guesses and emulation. If you have enough ‘as if it were in a brain’ components operating at different scales, it’s plausible. You can’t make three circuits conscious, but three billion circuits connected in a certain way might appear to process information as if they were conscious. If the ‘as if’ is convincing enough, you could argue they’re probably conscious. But let’s be clear: nothing being sold as AI or machine learning right now is conscious. Machine learning involves networks and circuits that can be adjusted based on repeated experiences in similar situations – that’s the essence of what it is currently.
As for AI, the term is a bit muddled because different groups – science fiction writers, futurists, engineers, computer scientists, and advertising agencies – all have their own interpretations. When it comes to defining AI, it really depends on which group’s understanding you’re referencing. But two things are particularly amazing: one, the fact that we’ve managed to create these networks of circuits capable of tuning themselves through feedback, which is mathematically straightforward but astounding in its simplicity and resemblance to the thought process. And there’s that quote, attributed to Einstein – real or not – about the universe being understandable. Then there’s another one about the effectiveness of mathematics in understanding the universe, which is also quite profound.
The concept that mathematics is remarkably effective in describing the universe has always been considered somewhat miraculous. Similarly, it’s a smaller but still significant wonder that simple circuits and their interconnections form the basis for learning capabilities. This is the first miraculous aspect.
The second miraculous aspect is consciousness itself. The rich information processing and association among specialized subsystems in the brain result in the sensation of existing in the world as conscious beings. Consciousness can be seen as a kind of technological marvel, an evolved technology. Our brains have developed to allow us to experience being in the world as ourselves.
Discussions about AI, depending on who is leading them, often blur these two marvels. They mix the small miracle of feedback circuits with the more ambitious hope of achieving the larger miracle of consciousness from these simpler foundations. In my opinion, this leap from simple circuitry to consciousness will likely happen reasonably soon.
[Recording End]
License
In-Sight Publishing by Scott Douglas Jacobsen is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License. Based on a work at www.in-sightpublishing.com.
Copyright
© Scott Douglas Jacobsen and In-Sight Publishing 2012-Present. Unauthorized use and/or duplication of this material without express and written permission from this site’s author and/or owner is strictly prohibited. Excerpts and links may be used, provided that full and clear credit is given to Scott Douglas Jacobsen and In-Sight Publishing with appropriate and specific direction to the original content. All interviewees and authors co-copyright their material and may disseminate for their independent purposes.
In-Sight Publishing 