Today Elon Musk unveiled to much fan-fair the latest iteration of Teslas ‘humanoid’ robot, Optimus. They’ll be mass produced, cost less than $20,000 each and be available in 3 – 5 years. Not a small feet, and I want one.
The rhetoric was nothing outside of the normal: “A future of abundance”, “a fundamental transformation of civilisation as we know it” and “a society in which robots [do] the work and people reaped the benefits “.
That robots and AI will become ever more abundant, and will change civilisation as we know it, I think is beyond doubt. But the important question remains: Will this be the utopian dream Musk proclaims, or follow the dystopian nightmares beloved by Hollywood?
For years there has been much talk about the impact of robots taking jobs. It’s a well trodden debate about skills, training and the ability to lift the repetitive, sometimes lower-skilled, workforce into the service economy. For me though the question is deeper. In a world where robots and AI do much of work and we ‘reap the benefits’ will we really be happier? And will this really level society or does it risk increasing the already gaping divide between the top and the bottom of the economic food chain?
Emotions
Do you really want to live in a world where machines do all the work?
As humans we are driven by the innate and fundamental impulses, baked into our primeval ‘old’ brain – Protection, nutrition, procreation (and as a subset social status). But what separates us from animals is our ‘new’ brain – the neocortex – our intelligence.
It’s this intelligence that created money as a way of measuring worth. It’s this intelligence that drives our complex hierarchies of social status – measured through social cohesion, original thought, innovation and emotional empathy, with money often used as the proxy measure for these. Where an animal would fight it out to find who’s top dog, we measure our status with these complex nuances. Consciousness alone is not enough to satisfy these. We need the satisfaction of constantly learning, of providing, of developing our brain to stimulate this intelagance. It’s through experience that we generate ideas and innovate. When the need to ‘do’ is replaced by automation where will our satisfaction and energy go? A rat presented with an endless supply of food will eat, sleep, procreate and die happy, but we have the propensity for boredom – the greatest sign of intelligence.
Our intelligence is most stark when emotions come into play. All mammals experience emotions. Emotions are routed in the hypothalamus, part of the old brain, but the massive size of our neocortex means emotions effect us in unique and more nuanced ways. Our neocortex takes the emotional signals from the hypothalamus and executes that emotion. It amps it up. But it also lets us consider emotions. We need that stimulation.
Mashies don’t yet understand emotion. Our current AI models can’t even really be considered intelligent. Neural networks, whilst smart, learn once and handle only one actively. They can identify hand writing OR they can catch a ball, but they can’t do both. The next big wave of AI innovation will surly lead us to AGI – Artificial General Intelligence – a more multi purpose model that can tackle a variety of problems, but it will likely still be emotionally nescient.
Music is perhaps the best example of why emotions matter. Animals and humans alike are affected by music, generating activity in our hypothalamus and triggering emotions. But with humans our enlarged neocortex allows us to respond to music in deeper more nuanced ways. When we listen to music we empathise and relate to the messaging within it (linguistic or not). The emotion in the music resonates with our mirror neurones and allows us to feel the passion conceived and felt by it’s creator. It’s this ability to generate musical empathy that makes us all so musically susceptible – Music can overwhelm us, resonate with our mood or fundamentally change our current mood and energy levels. Have you ever tried watching a thriller or animal documentary with just subtitles? It’s generic and boarding – you don’t feel the same emotion. But as soon as you turn the sound back on, and the music is added, queue the waterworks. The extremely complex ability to relate myriad sensations to varied context is (currently) unique to humans. Within computer science this is called the ‘Knowledge Problem’.
AI can statistically analysis music, feed it into a neural network, and ‘learn’ what makes an appealing sequence of notes. But it won’t be innovative. It won’t be unique. It won‘t come from deep down, painful emotions embedded in the human soul.
The same can be said for craftsmanship or the emotional resonance that’s so important for good healthcare.
For AGI to truly become part of our lives we have to consider how it will effect our emotions, what emotions we should give it (a much harder task than giving it intelligence) and how we will keep our intelligence stimulated so we remain happy and fulfilled. The shift will have to not only be in society but also in our perceptions of self worth.
Economics
The second question – can these advances really level society – again boils down to economics, and as ever it’s a lot easier to make money with money already behind you.
Developing robots and AI / AGI will costs billions, and we are far from there yet. The concepts required to move from AI to AGI haven’t even been conceptualised yet, and indeed some think it might not be possible to truly represent consciousness in a machine (I don’t agree). Rightly, the inventors and investors who have placed their capital and energy into these high-risk projects will expect a high reward. So the question becomes: How will they seek to recoup their investment, and how long will it take for the power of AGI to become truly accessible to everyone? Will these investors seek to monetise their investments by leveraging them them selves (using the intellectual power advantage they will deliver) or by producing for the mass market straight out of the gate? It took roughly 60 years to go from singly-purpose ’computing’ devices to the universal Turing machines we all have in our pockets now. Once we get to AGI will it progress at a dramatically faster pace, but will it that pace be throttled for self interest?
The next leaver is education. When it comes to re-skilling the rich again have an advantage. They can access better education for their children, afford continues education for them selves, and buy in highly skilled workforces. In the early years of AI / AGI / robots education is going to be what allows people to leverage advances faster, and in doing so capitalise on the advantage to grow their personal wealth. Once AI / AGI / robots become the norm, again the rich will likely benefit from their existing general levels of higher education – they are generally either less exposed to the manual and lower skilled work that will be swiftly replaced.
The last economic leaver is access to working capital. The rich have better access to both networking and the working capital to execute their ideas and dreams – that’s not to say their ideas and dreams are any more valuable, just that they have a better chance of becoming reality with low risk capital behind them (if the idea flops, the money lost will likely not impact their lifestyle noticeably, making them less risk adverse). They’ll also be better placed to leverage these opportunities to increase existing business margins. In the inevitable years before AGI reaches mass availability, the rich will be able to move more quickly, widening the economic catechism we see already.
It’s also worth a passing note that the wealthy have far grater access to intellectual and stimulating pastimes. In a world where we don’t all have to work in the conventional manner that matters.
Musk’s $20,000 robots will not only be far from the AGI mark in the next five years, but are also way outside the price bracket of your average want-to-be entrepreneur.
Massive Upside
It doesn’t have to all be doom and gloom. The upsides are enormous. Health care costs will plummet. Opportunities we can’t even dream of will become reality. Ultimately machines will become extensions of our own consciousness and allow fundamental shifts in how we see the world.
The internet provided an opportunity to level up – anyone can launch a website now and access a global market. AGI will eventually achieve similar, and likely greater feats. What will hold us back is fairer access to high quality education. We need to re-think how we distribute value within our societies. Loss aversion is baked into us all – it’s an innate behaviour – part of our primeval need to provide and secure protection and nutrition, and is rarely stronger than with money. To allow this technology to really help create a more equal society, with truly equal opportunities, we need to re-consider how we judge and value our economy – possibly even money as the base of our economy. This is truly a moonshot that is to early to really ponder now. But education is something we can and must focus our energy’s on changing.
Notes: There are two books that I would highly recommend to anyone thinking thing in this space. The first is A Thousand Brains, which poses fascinating new concepts for intelligence and the brains mechanics, and how this may apply to AGI. The second is How Emotions are Made, which challenge the long perceived views of our inner emotional processes.