Empowerment Versus Enfeeblement
Part 6 of Artificial General Intelligence (And Superintelligence) And How To Survive It
Even a rapidly unfolding emergence of superintelligence could take place over months or years, and thus in its early stages still be shaped profoundly by how we marshal and deploy our resources – human, compute, financial, technical, energy and so forth.
Which brings us directly to the question of enfeeblement, because confronting a mercurial job market and the need to figuratively and even literally “upgrade humans” are two obvious ways to improve our capacity to control, guide and partner with AI in the era before full ASI becomes possible and even commonplace.
Simply put, we want Empowerment, not Enfeeblement, and how we finesse disempowering trends will be critical to the future we are collectively creating.
America’s tech companies have famously fired a few hundred-thousand employees in recent months.
Human costs and the loss of immense human capital weight heavily on the minds of any strategists considering the loss of so many talented, highly educated knowledge workers.
But faced with such disruptions, we should also be looking at opportunities.
In this case, we have entered a world in which there is a grave shortage of skilled cybersecurity experts, innumerable holes in our collective cyber defenses, legacy code – and simply bad code – weighing on businesses and governments alike and there is a dearth of expertise available both to assess AI safety but also to leverage it for companies, organizations and entrepreneurs who could find it invaluable if only they had the means to use it.
The potential to redirect these human resources should not be underestimated.
How to do so will by its nature be complex, because we are not looking at a static picture. People left big tech and went looking for other jobs at the same time as thousands of their colleagues. Some of them inevitably found jobs, though perhaps not to their liking in both compensation and fully utilizing their skills.
The reality, though, is we have moved from a scarcity of programmers to a glut, and given our imminent needs, the free market is not moving fast enough to redeploy them.
Ideally, countries with a sudden overload of skilled programmers should take a hard look at their most critical strategic concerns, prioritize them, and then turn this embarrassment of riches into a reserve of talent being directed effectively and at scale to address them.
These are sizable, but let’s break down a few to get a sense of them.
In cyber, we have the fascinating paradox that the US Federal government has already acknowledged using AI to automate cybersecurity, a state of affairs which will not only never stop, but only intensify as we move forward. Humans are doubtless still in the loop in many decisions, but the speed and complexity which are the hallmarks of modern computation paired with artificial intelligence make this fusion inevitable.
AI cybersecurity systems, however, are apt to be incredibly well-guarded secrets, so you won’t be downloading the most advanced AI cyber tools onto your laptop anytime soon, even if it were capable of running them. But while we may keep our best innovations very close to the chest, cyber isn’t about protecting just a few big systems for a few wealthy, powerful organizations.
It's about protecting all of us, the network of systems in our lives, and even our networked devices, down to the humblest of microprocessors. If your business crashes without connectivity or your phone can be hacked and used to spy on you through its microphone and camera, suddenly even the most-common devices become vectors of attack.
A quarter-million programmers of considerable skill would be a formidable force, but given many of them have new jobs and even new careers, we might not be able to recruit such numbers even if we budgeted for them.
But imagine 100,000 such programmers, averaging – for the sake of the argument – $150,000-a-year and – given the time needed to hire them – working for an average of 6 months in this fiscal year. That would cost about $7.5 billion through the end of the year, and $15 billion for the whole of the next one.
We can quibble about the numbers, but suddenly a literal army of programmers you can throw at multiple key projects at will seems far more plausible, even at twice that price.
Mobilizing such a force may seem like a blunt instrument – and it is, especially if it’s your sole instrument.
You will want tools which enable the mustering of vast human capital where appropriate, but also the flexibility to avoid a one-size-fits-all solution or even the assumption you’ll automatically know where best to deploy every new expert at your disposal.
So consider, in this case, we have an immense influx of professional programmers into the job market, now fiercely competing with each other, while we also have a number of critical goals which could use their expertise.
But not every programmer is an expert, or even particularly well versed, in cybersecurity. Other, more exotic skills such as AI alignment or AI red teaming will be even more sparse.
Yet you also have professionals – many of them elite professionals – with years of education and years of experience.
How then best to turn this asset to our highest goals, while also fulfilling the highest aspirations of businesses, governments and nonprofits throughout our nation? A question every country should be asking.
What if we began by extending an opportunity to key professional sectors affected by AI displacement, as a pilot program?
What if the program not only paid for retraining or new education, but paid the displaced professional a significant – though far from exorbitant – salary for retraining or reeducating to fill a new niche with their refined skills?
If an IT specialist can pick up the necessary abilities to handle basic cybersecurity tasks in several weeks and that is all they aspire to, well and good. So long as we know they are taking the courses, they can receive their pay. If retraining is essentially a semester or two of further study, again, they can take their coursework, reap their benefits, and potentially have a job offer at the end of it, though not a mandatory one.
This could go as far as two years of paid study for people adding a Masters or PhD to their educational portfolio, or expanding their skills in completely new direction such as biotech applications or research in human-AI teaming.
Initially, more extreme benefits packages likely will come with strings attached, such as a commitment to serve in the government or high-priority sectors. But as those sectors will include a great deal of private industry we need to secure, or AI projects we need to further, the educational assistance does not mean a definite government career and certainly does not mean a de facto oath of poverty.
Creating a highly talented IT reserve inside the government, though, will provide us with a reservoir of skills which can be shifted to many pressing needs. Cyber and AI, to be sure. Assisting in our research into and management of other emerging and critical technologies, of course.
But how pressing are the overall IT needs in the military? In intel? In the national labs, NSA, DARPA and its descendants such as IARPA and ARPA-E?
How much bad code and outdated software do we have dragging down the entire government?
Nearly everything from fusion research to biosecurity could use an IT upgrade. An army of not just cyber warriors, but IT builders, could help us to transform, and do so with speed.
But the Federal government is not the only organization with pressing IT needs. Cyber, as noted, impacts every state and local government, and virtually every business and nonprofit. Everyone struggles with bad code, or at least less-than-superlative software.
And while there are many heroic efforts to be undertaken, how much of our cybersecurity would be enhanced if we simply did basic audits of glaring errors, even as extreme as a lack of a firewall or antivirus software on devices? Using heavily compromised antivirus programs, such as the alleged issues with Kaspersky products?
The above scenario is not a policy prescription, but rather a starting point.
If a mix of incentives, tax breaks and contracts could achieve the same goals even more efficiently, we could certainly do it that way instead.
The key point is to ask whether we have an excess supply of strategically vital talent, and existential threats to throw them at. And whether we will therefore use that talent pool, and how.
Remember also, so far we’re only discussing the circumstances involving programming talent in one country, America, and any advanced, wealthy economies facing a similar deluge of underemployed software engineers. Not only will the details differ radically between nations, some will have different yet viable solutions, and others will not be dealing with this displacement of talent at all, particularly if they didn’t have it in the first place.
Again, it’s a place to start, not an endpoint of our discussions.
We want to consider the programmers because that issue is already upon us in the West, and paradoxically, they are among those best suited to helping us make this leap to a new AI age.
But they will not be the last.
What happens if we truly cure all forms of cancer? Such a feat would be a blessing, yes.
But what happens to the oncologists?
More on this later, but here we have another example of gifted people whose base professional expertise is of immense value, but who might be largely lost to us in a sudden shift in our medical knowledge.
We don’t need to lose any of them.
Here we come to another emerging technology, which is more of a catchall description than a field…
Human Enhancement.
Human enhancement is literally anything which enhances the function of the body, the mind or both.
By definition, it is an almost ludicrously broad area of study.
But it is also an example of how humans, at least in the early stages, will be able to keep up with the machines.
How so? And what does this have to do with our doctors?
An excellent question. Because our first example of human enhancement isn’t biotech or educational innovations.
It’s electronic. With the increasingly prevalence of affordable smartphones in the developing world, paired with the growing accessibility of electricity via the grid or renewables, millions of people have gone from no computer access to carrying around a miniature supercomputer with an Internet connection to most of the world’s knowledge.
All in their hand, or in their pocket.
The moment someone asks whether human processing power can possibly keep up with AI, remember…
If AI’s compute goes to infinity, then no, we can’t.
But those smartphones added a powerful computer’s worth of capacity to each of those humans.
And they could be further upgraded with their own AI agents, and probably will be if they haven’t already. To the extent that AI-empowered systems are filtering into the hands of the public, we already see human-AI teaming.
But it’s not just the software, or augmented reality hardware, which promises to upgrade us.
While AI in human enhancement is too vast a subject to delve into just now, much of this work is done in the biotech sphere – enhanced diet and exercise, no doubt, but also nootropics, epigenetics, genetic engineering and cybernetics. More exotic work includes biofeedback, transcranial direct current stimulation (tDCS), cranial electrostimulation (CES), and sensory-deprivation tanks.
Any existing prodigy or genius significantly enhanced using viable methods and technology would become a potent if not unique factor in their field. A Leonardo da Vinci, Srinivasa Ramanujan or William Shakespeare thus augmented would become effectively superhuman.
There are many people in the modern world already brilliant in their field, or some specialized sliver thereof.
What happens if we noticeably enhance these individuals and they work together, further assisted by human-AI teaming?
Some methods already exist, and synergize well with other enhancement techniques, but a key thing to remember about human enhancement is how personalized it is. More scientists and doctors would be welcome in this research, but so would medical professionals capable of working with individuals in augmenting their skills, compensating for weaknesses, and making sure each regimen was not only ideal for each recipient, but avoiding pitfalls – such as unforeseen drug interactions between prescription pharmaceuticals and nootropic supplements.
An even more basic, therapeutic form of human enhancement is to increase medical access and affordability for underserved communities on the one hand, and to improve general nutrition, fitness and health on the other. Merely optimizing human bodies and minds throughout the population has numerous beneficial follow-on effects.
So excess doctors could easily be put to work in other branches of medicine, or in the development and application of biotech means of enhancing humans. Places of employment could vary from emerging technology research labs to rural hospitals.
This may seem extreme, but first we need to think outside the box, and second, as with improving the population’s medical services or fixing fundamental cybersecurity flaws, we have to realize a lot of very basic work could prove the foundation for critical initiatives in managing the effects of AI.
Addressing the displacement of workers in strategic professions will also acclimate us to the idea of helping the larger population adapt to rapid change, even if keeping their specific jobs in place may be less the objective than helping them move to a better option.
The key to worker displacement is that in helping others we can also help ourselves.
And nothing is foreordained in how we respond.
Courage and vision are more than welcome, but forethought lets us plan for eventualities, and consider innovative paths before we become trapped in a labyrinth of unexpected consequences, where all the walls are relentlessly closing in.