Experts have said the singularity, the inevitable time when AI outsmarts humans, is pinpointed at around 2040 to 2045. On the other hand, hybrid thinking (when we merge brain and machine) will be available in around 20 years. (20 years sound optimistic, considering the engineering challenges and the bandwidth problem. Plus, issues of scaling, cost, and politics may push it a few years back, along other factors discussed later.)
One side, led by Elon Musk, is summarized bluntly by his statement that we either merge with AI or be taken over in the revolution. In this sense, cognitive enhancement seems like our last stand against becoming the pets of AI. But this far-fetched vision seems only supported by two private companies right now – Kernel and Neuralink, both individually funded from two already-wealthy men who made profits off e-payment companies… Research in this area, however, is garnering momentum but nevertheless restrained – coined as “mad science” in this article on research being done by DARPA.
The other side, much less sympathetic to humanistic views, is that the singularity will occur before such hybrid thinking is commonplace, and the AI will take over as predicted.
The bystanders, basically all the big data analytics companies, seem not to care, as long as AI keeps them competitive for as long as possible (those who don’t adopt AI will die out anyways, so this is inevitable by market Darwinism).
This post is the accumulation of all the research I’ve done, how I think the way events will turn out, and what aspiring visionaries (including me!) can invest their energy in at each step.
- AI Revolution (Present-2030)
The current deep learning craze will continue. Specifically, deep learning products will come to market, including self-driving cars, Amazon delivery drones, personal assistants, apps, etc. Most corporations will require data analytics (“data is the new raw material”) to survive, so demand for AI people will continue to rise. By 2030, the revolution of Artificial Narrow Intelligence will be over, as AI (combined with robotics) will be in every aspect of our lives (Japan is ahead of the game). As the Artificial General Intelligence (AGI) revolution matures, there’ll be development of learning algorithms beyond neural networks (from reading answers like this on Quora, current algorithms cannot create AGI; a bottom-up approach must be done). AGI will begin taking on management positions overlooking all sorts of operations (majority of GDP will be created by AI as well as financial transactions, considering >50% bonds traded just last year were done by computer). Near 2030, unemployment will be at a significant problem, not counting all the discouraged workers. Protests for change will become serious.
- Beginning of Learning Revolution (2030-2040)
Around this point, drastic structural employment will be serious enough urgent summits will be held to discuss measures to appease the damages done by the previous 20 years. In a 21st century reincarnation of the New Deal, government programs will generate temporary but futile solutions for workers. Then, heavy funding will be given to the education industry to retrain workers and the next generation, hence the start of the Learning Revolution. At this point, the brain-machine interface (BMI) bandwidths will be high enough for employable neurotechnology. This is when the Silicon Valleys of the world will (as it is today) begin making schools obsolete and transferring educational material to non-invasive implants.
Such an implant will not “upload” skills to your brain, but instead will try to speed up learning. They serve as targeted neuron stimulation that can activate specific neurons (or “cliques” of neurons, see here). Firing of all action potentials will be recorded then sent to the cloud to be compared to that of others in the database. Naturally, the top performers’ (i.e. chess grandmasters) logs will serve as templates for comparison to regarding the brain activity of the user. A feedback loop (back propagation) then occurs, the goal being to transfer the timed brain activity of the signal to the user when doing identical tasks. This is essentially accelerated rewiring of neuroplasticity (which occurs naturally through practice of a certain skill).
In layman terms, the hope is that, by at least wiring your neurons to fire the same way as Magnus Carlsen’s during a chess match, you’ll see results much faster along your journey than otherwise.
I have yet to see an expert echo the concept of the implant I describe above, but the closest step in this direction I’ve seen is a research done in the labs of U-Penn which uses such stimulation to increase memory. Though the enhancement is only at 15%, quality software should scale it proportionally by bandwidth, so it may be a 100-500% increase for general learning capabilities by 2030.
This will indeed be revolutionary because it renders traditional low-bandwidth instructional methods obsolete (teacher’s brain -> teacher’s mouth -> student’s ear -> student’s brain). Instead, this will allow expert’s brain -> student’s brain instruction. Classrooms and textbooks will be anachronistic systems that will be eradicated entirely, which will cause protests that delay this transition by upwards to a decade as people debate the future of school and children.
While this implant is revolutionary, it will likely be costly and ahead-of-its-time, so only a small fraction of the population will own it. This will likely create an even stronger meritocracy, which only aggravates the damage of the AI revolution. Instead of being a last stand against AI, I predict it will backfire and create more inequality.
Another foreseeable event is a “Manhattan project” described in this alarming talk by Bostrom for founding fathers to grind out a Constitution for the release of AI (security, goals, preventative measures, etc.) This will likely be successful; otherwise, government will ban the release of AGI. Government policies will be created as well, one being the popularly endorsed Universal Basic Income.
- Singularity occurs (2040)
AGI will be complete here, and quoting the words of experts of Tim Urban (waitbutwhy.com), Sam Harris, it’s intelligence will likely “zip” right past us as intelligence is exponential (AGI, not us, will create Artificial Super Intelligence, the event horizon of our history of innovation). AGI will begin taking over all jobs in the private sector in the US, including R&D and scientific research. ASI will be created to optimize humanity and govern the AGIs (some formula for happiness under a lot of conditions; for a cautionary dystopian story, read this). We will have finished our last invention.
There will likely be a divide between the “Old World” and the “New World,” as the last of modern society finishes the transition into this “egalitarian” world described by Musk.
- Cognitive enhancement affordable to the masses (2040-2050)
By the time we’re in this “New World,” the demand for the learning implants described above will skyrocket. But the biggest market will likely be for virtual realities. The learning implants will amass the collective knowledge of mankind (and discoveries made by AI), and can be accessible to everyone at a cheap cost (think automated googling for anything you want to know). To prevent memory overload and greed, there will be a certain amount of “credit” each person can use to “shop” for knowledge they want to know under a budget. For example, you will be accessible to develop the musicality of contemporary Mozart or the understanding of Einstein, but not both. Once you commit to a field, your job is to try to extend the limits of progress, which adds to the collective knowledge of mankind for the next generation to access.
- Result (2050-Beyond)
This efficient system is beautiful because it will not only optimize human progress via specialization but provide “every single person” with “a sense of purpose.”
Outside of your “job,” you can acquire any life hack to use (how to tie a tie? You will know before you finish asking the question).
Entertainment will occur mostly in the virtual world, but will be restricted to after work hours. Intentions of over-indulgence can be squelched in the brain as they form. In this sense, virtual world hours are the “currency” that people will speak of.
My belief is that a prerequisite to utopia is the elimination of competition that pertains to scarcity of resources that directly relate to well-being (ancient times: food, shelter; now: reputation, power, money). This system will provide everyone with rights to an ample life, then layering on top healthy competition with others to be the best in the path you chose for yourself, under the purpose of advancing humanity.
I acknowledge the specifics in this post are unlikely to be how actual events turn out (and I’ll adjust accordingly), but I’ve made it my mission to advance this vision forward both in research and entrepreneurship. If I can play the role of steering the world such that it turns out the way I envision, I’ll be very, very happy.
表裏一体 