Generative AI and the Dark Forest

Welcome to an infinite deluge of digital crud, sneaky scams, and exploitative fakes - plus some cool new videos!

For the last few days, I have been reflecting on the increasingly glaring problems with Generative AI as it proliferates across the Internet. Like a fungus or cancer, it keeps growing, devouring everything in its path.

I remain curious about AI and continue to use and experiment with it in various ways. As one example, we have been making promotional videos for The Elevator (subscribe to our newsletter), our think tank exploring consciousness (these videos are produced by Guy James):

I also recognize that AI could exponentially accelerate scientific breakthroughs that might help us find as-of-yet unknown solutions to the ecological emergency. For instance, Google’s Deep Mind developed GNOME (Graph Networks for Materials Exploration), which rapidly discovered millions of (hypothetical) new materials. As Google’s AI told me: “These materials include graphene-like superconductors and potential lithium ion materials that could be used in batteries.” The International Atomic Energy Agency is using AI to accelerate the development of nuclear fusion. Fusion is our “moon shot” to maintain anything like this global industrial civilization as we run low on fossil fuels (which, you may have heard, also cause runaway heating).

But it seems ever-more likely that the negatives of AI will far outweigh the positives. AI doesn’t seem helpful or conducive to human flourishing and freedom, except for those in the small coterie who will become very wealthy in the short term. The dangers of AI for military misuse, government and corporate surveillance, behavioral manipulation, and deeper levels of mind control are staggering. I explored some of these in my recent podcast with Warren Neidich, “Your Brain on Cognitive Capitalism”.

AI cuts deep into the revenue of already struggling newspapers and magazines. Chat-GPT and other AIs respond to user queries without users needing to go to the original news sources, thus reducing “clicks”. The integration of AI into search engines amplifies this effect. This is already causing mass lay-offs in journalism. The New Yorker called AI an “extinction-level event” for journalism and media. The collapse of local investigative journalism is extremely beneficial for Right Wing politicians and their corporate backers, who can exploit and extract without oversight.

(This is a bit of a digression, but if you have time, I recommend reading “A Secretive Hedge Fund is Gutting News Rooms”, from The Atlantic a few years back. A group of investors buy up US newspapers, treating them like oil wells to be siphoned for as much profit as possible, strip-mined for assets, and then abandoned. For these predators, McCay Coppins writes, “Newspapers are financial assets and nothing more—numbers to be rearranged on spreadsheets until they produce the maximum returns for investors.” This has negative impacts on local communities: “Take away the newsroom packed with meddling reporters, and a city loses a crucial layer of accountability.” )

It is difficult to envision a scenario in which we are able to turn AI off, at this point, or even restrain it. Of course, this is not impossible nor inconceivable. But how would we do it? As Erik Hoel notes in “Here lies the internet, murdered by generative AI”, this falls under the famous “Tragedy of the Commons” problem, where rational actors drive a system toward destruction through competition. For example if US companies don’t achieve AGI first, we will cede this to the Chinese or the Russians, who would gain a massive advantage.

As with many situations in our society, it is much easier to marinate in the problem than it is to envision, let alone enact, a meaningful response. With generative AI, the situation already seems quite overwhelming.

Hoel, a novelist and neuroscientist, explores the massive influx of derivative trash, fake content, and pornographic deep fakes already unleashed by AI, like toxic sludge clogging up the digital superhighway. He writes:

Now that generative AI has dropped the cost of producing bullshit to near zero, we see clearly the future of the internet: a garbage dump. Google search? They often lead with fake AI-generated images amid the real things. Post on Twitter? Get replies from bots selling porn. But that’s just the obvious stuff. Look closely at the replies to any trending tweet and you’ll find dozens of AI-written summaries in response, cheery Wikipedia-style repeats of the original post, all just to farm engagement. AI models on Instagram accumulate hundreds of thousands of subscribers and people openly shill their services for creating them. AI musicians fill up YouTube and Spotify. Scientific papers are being AI-generated. AI images mix into historical research. This isn’t mentioning the personal impact too: from now on, every single woman who is a public figure will have to deal with the fact that deepfake porn of her is likely to be made. That’s insane.

Hoel points to AI-generated “workbooks” sold on Amazon, pillaging from one of his books, as just one example of the endless scams already proliferating. An article in this week’s New Yorker, “The Terrifying A.I. Scam That Uses Your Loved One’s Voice”, looks at how thieves are using voice cloning technology to imitate the distressed voice of a member of a family. They call another family member and demand an immediate ransom via Venmo, or their relative will be executed, gangland style.

As for the plusses, many of them seem a bit evanescent when you stop to reflect on them. For example, I admit to being intrigued by AI video. I’ve got two screenplays languishing on my hard drive which I think could be amazing, groundbreaking films. But I lacked the necessary pedigree or professional connections to get them taken seriously by Hollywood. If I can use AI to make them into actual films in the next few years for close-to-zero budget, I will do so.

Last week (I think it was — time moves so fast in the world of AI it is exhausting to keep up), Open AI premiered a new text-to-video technology, Sora, which can manufacture up to a minute of AI video based on a text prompt. Some of these videos look startlingly real, but many also contain bizarre flaws. While this technology will keep improving, it may also hit a ceiling in terms of what it can do.

Also, giving everyone the capacity to produce massive amounts of imagery and video may, in the end, induce apathy instead of inspiring any kind of creative renaissance. It is kind of amazing, when you think about it, that giving everyone a super advanced camera in their Smart Phone has not led to any kind of creative avant-garde in contemporary film-making. If anything, even independent films seem increasingly formulaic, except for a few very rare exceptions.

It’s extraordinary that Open AI — the biggest culprit in opening the Generative AI spigot — didn’t think through the consequences of releasing this infinite deluge of trash.