AI’s Reputation Problem
Photo by Lyman Hansel Gerona on Unsplash
If you know me, then you know that when it comes to AI, I tend to take a pretty bearish stance.
So much of what you hear when people and companies talk about AI is what it can do, but it’s done in a sort of, “coming soon to a theatre near you,” breathless kind of way. It’s certainly a fun tool to play around with, but the problems that underscore the technology, nor how over-hyped it is, shouldn’t be ignored either.
One of the main reasons I’m not so hot towards this new technology is that Tech as an industry is in a reactionary valley right now. Companies are desperately trying to reclaim some of the lustre they lost during the mass layoffs that took place in 2022 and 2023. Prior, money was flowing freely, so much in fact that it has a name - ZIRP - and is partially responsible for the rise (and subsequent crash) of the first two web3 pillars; crypto and NFTs (Zeke Faux's excellent book, "Number Go Up," documents this - I highly recommend reading it).
Speaking of which, the second reason I’m not sold on AI yet is that right now, it’s just a bunch of web3 bros (and web3 wannabes) pumping the technology because, with ZIRP gone, it’s where the money is. What’s remained is the same breathless zealotry for a revolutionary new technology that will change the world. Sound familiar? It’s why every company has pivoted into AI, and why every start-up is an AI start-up (and why sometimes AI is literally just 1,000 underpaid workers in India).
Comic by Benjamin Schwartz at The New Yorker
Let’s look at an example in the form of Marc Andreeson (cofounder of Netscape and later the VC firm, a16z). Andreeson was one of the earliest investors into the crypto and NFT craze Ponzi schemes, and when that came and went, he pivoted into… yep, you guessed it, AI. He pivoted so hard into AI that he spent the time to write a 5,000-word “manifesto” that sure says some things. From the New York Times:
"… the rambling and often contradictory manifesto has the pathos of the Unabomber manifesto but lacks the ideological coherency. It rails against centralized systems of government (communism in particular, though it’s unclear where Mr. Andreessen may have ever encountered communism in his decades of living and working in Silicon Valley) while advocating that technologists do the central planning and govern the future of humanity."
What this NYT article fails to capture, however, is just how uncomfortably close Andreeson’s words are to the ideals of eugenics (he quotes noted Italian fascist, Filippo Tommaso Marinetti, early on as he envisions a race of technologically augmented “supermen,” and “conquerors”), which seems to be a bit of a running theme among the TESCREAL crowd.
The final reason (for now, at least) that I don’t think AI is what we’re being
told it is/can be is that the sheer amount of data required to keep the machines learning will outstrip the available supply within two years. This means that AI companies are leaning into the idea of generating their own synthetic data in order to continue training their systems - an idea that should horrify you if you’ve ever witnessed the kinds of hallucinations AI is already prone to with non-synthetic data.
One thing that dweebs like Andreeson are right about is we do need technology, and that we’re unlikely to solve many of the world’s problems without it. But if AI is allowed to supplant real human creation and curation, then we might be heading into the Dead Internet a lot faster than we’d care to admit
