Maybe, just maybe, the problem isn’t with AI, but with how we’re designing it, and perhaps more importantly, how we’re using it.
We are told every day that it’s too late to stop AI. That’s bad, because there is a lot that’s wrong with today’s AI (including some major critiques). Above all, it’s just not true at all that it is too late to stop AI.
Let’s start with why current AI is bad. There’s no shortage of reasons to oppose the current rise of AI, but if you think that those reasons include human extinction or mass unemployment, think again. Thoughts along those lines are just mass distractions. AI could exterminate humans only if we were so unbelievably stupid as to seriously deserve it. Similarly, concrete fears that AI shall destroy all jobs should take second place to questions like:
- If AI can really already do all the jobs that humans do, aren’t many of those jobs (some say as much as 30% of the economy) useless or otherwise unfit for humans to begin with?
- If, instead, AI increased every worker’s efficiency so much to keep everybody employed, but with made-up, life-sucking jobs… would that be progress?
The real, main problems with current AI include, in no particular order:
- Current AI brings real dotage and isolation. It has been said that “AI can literally give us access to spaces that we, on our own, acting as or being human, cannot discover and cannot access.” That’s true, but then again so could LSD, which if nothing else had the merit that nobody could trip on acid for almost five hours per day, every day, since seventh or eighth grade.
Current AI, instead, produces so much fake content, so fast, that meaningful digital conversations among humans become almost impossible. Human taste for what’s good disappears and most of us get dumber, from youngsters trying to date, to researchers and software developers — the very people who can benefit from harnessing AI.
- Current AI is only a figurehead, or an extension, of the really few people that control it. That means we already know that AI as we know it is toxic, because it’s been around for centuries, in the guise of corporations that people, not fate, made immortal and equal to humans. Current AI is just another corporation.
- Current AI is a perfect smoke screen to mess with democracy and freedom. More exactly, current AI is great for “manufacturing excuses and replacing politics, automating not paperwork but democratic decision-making… If you want a labor force, a regulatory bureaucracy, or accountability to disappear, you simply say, ‘AI can do it.'”
- AI is really, really hard to make open, and this alone should be enough of a reason to get as little AI as possible.
- Current AI just won’t deliver, in large part because, says MIT professor Acemoglu, it’s being “applied to the wrong types of problems in the wrong professions” (more on this below).
- Current AI is childish. It is just the latest incarnation at all levels, from energy and water demands to avalanches of fake security reports, of the “more and faster is always better” attitude that balanced people outgrow with age.
Reasons Why Current AI Is Not Unavoidable
We are told that stopping current AI would be not just impossible, but really, deeply wrong. That’s simply false, and here’s why:
- AI is nothing we are obliged to create, preserve or respect. No way. This is not nostalgia for human exceptionalism. It’s self-preservation as a species, which is the most intelligent thing there is. Even if it were possible to create real artificial intelligence, we humans are under no obligation whatsoever to let happen, begin, or survive anything that threatens or vilifies us. And I say “anything” because…
- AI is a mere something, not a “someone.” The fact that some immature kids with socialization problems desperately need something else to talk
to because they can’t stand humans doesn’t make AI necessary, and surely does not give AI any human-like status. - AI is not strong, nor independent. AI is not COVID-19. AI is just software: something that ceases to exist the moment a child unplugs the computer it runs on.
- Big Tech AI is a sinkhole waiting to open, and its puppetmasters know it. DeepSeek demonstrated that it’s possible to match Big Tech AI with a tenth of the computing power that we previously thought it needed. That is, there may be no need to invest in new data centers to solve some real problems with AI, or exactly because DeepSeek is cheaper, create other problems faster.
That’s why many Big Tech reactions to DeepSeek sound like “Don’t worry, investors! [An] obscure economics concept shows a profitable path forward even though our competitors are drastically undercutting us.”
Indeed, the bigger and most successful AI products — run by a few, unaccountable American corporations — seem to be the lobbying to kill or rewrite AI related legislation worldwide, which they do exactly because their position is really fragile.
What’s Unavoidable Is Rebuilding AI on Reality
We humans were not made to be real-time, mentally bulimic beings, and we certainly were not made to help any of us to become god. We don’t need general purpose bots that can fake everything from luxury resorts in Gaza to intimate relationships, school essays and looming “singularities.” The AI we really need, the sooner the better, is reality-based AI that:
- Is as small and slow as possible at all levels, from resource consumption to intrusiveness. This alone would spare a huge amount of problems.
- Is “completely disabled without a more human, neurotypical partner.”
- (Quoting Acemoglu, again) provides really reliable, actually useful assistants for “educators, healthcare professionals, electricians, plumbers and
other craft workers.” - Solves real, urgent problems of typical humans like, to make just a few examples:
- Helping doctors in the right ways.
- Standardizing all the world’s boxes.
- Rewriting laws and regulations that humans, and humans only would validate and apply, as I described here.
Recommended reading: “Why Bad AI Is Here to Stay“, which explains why what many may call “Bad AI” could still be quite useful instead.
Marco Fioretti is an aspiring polymath and idealist without illusions based in Rome, Italy. Marco met Linux, Free as in Freedom Software, and the Web pre-1.0 back in the ’90s while working as an ASIC/FPGA designer in Italy, Sweden, and Silicon Valley. This led to tech writing, including but not limited to hundreds of Free/Open Source tutorials. Over time, this odd combination of experiences has made Marco think way too much about the intersection of tech, ethics, and common sense, turning him into an independent scholar of “Human/digital studies” who yearns for a world with less, but much better, much more open and much more sensible tech than we have today.
Be First to Comment