Factual Dispatch #65 - Drowning in Electric Sheep

Do Androids Dream of Electric Sheep, the Philip K. Dick novel that Blade Runner and its successor 2049, were based on, has been on my mind lately. Not because of the weird fashion, the overpolicing of the poor, or the climate change ruined planet, but how synthetic beings interact with each other. Twitter is mostly bots or AI-spam these days, Facebook is hollow shell filled with automated reposts of memes & pirated movie clips. LinkedIn is where the toxic positivity could still be generated human acolytes of late-stage capitalism or AI, but it doesn’t really matter.

AI follies run the gamut from the tragic to the comical to “Black Mirror but Stupid”: Deepfake scam videos, deepfake & synthetic pr0n videos, lawyers citing legal precedent that doesn’t exist, researchers listing fake citations/publications in their bibliography, Amazon product listings being infested with AI text, Google Ads using AI-driven translation software to hide the offshore location of its support reps, voters getting completely wrong info about presidential candidates from chatbots, and magazines replacing their entire writing/editorial teams with AI, leading to disastrous results. Researchers have been using synthetic text for years before ChatGPT, but now no one is reading the text that is output. Doesn’t matter if it’s in an academic abstract, a legal brief, or a sponsored post:

Why? Becaust AI is checking the output for us right? That’s what they/we keep getting wrong. They aren’t “alive,” nor are they “thinking.” LLMs like ChatGPT are “thinking” or “remembering” as much as the FB chatbot your florist uses, thinks. AI systems can’t recall or update themselves like traditional databases or Google’s search engine. A research paper titled Transmission Versus Truth, Imitation Versus Innovation: What Children Can Do That Large Language and Language-and-Vision Models Cannot (Yet) reminds us that ChatGPT can’t do on its best day, what a fussy, hungry 4 year old can do It can’t come up with a novel solution it hasn’t seen before. It can’t synthesize, innovate, or recognize. The abstract of the article:

This is why projects marketed as “100% AI” keep turning into Wizard of Oz or Mechanical Turk scenarios. Ed Zitron unpacks an egregious episode of this in Empty Laughter, where Will Sasso and Chad Kultgen pretended they built a George Carlin AI that they had an interview with. After the threat of lawsuit from the Carlin family, they admitted it was them who wrote up the script the AI pretended to speak.

When you’re expecting human-created writing and get AI, it’s as bewildering as when you’re expecting AI and get a human. That’s the issue, not that Sasso can’t write, but that we were grading his writing on a synthetic curve. The reverse is also the case, where 80% of influencer accounts don’t designate ads as paid content. It’s not that we don’t understand ads are an integral part of the web, but when they aren’t even disclosed, it is impossible to differentiate high quality ads from “good content.” It took the FTC stepping in to force Instagram influencers to label their sponsored content as such. On Twitter, it’s still going on. Even worse, Musk spiked the revenue share that MrBeast got, to showcase the “pivot-to-video” he did to make up for the continuous loss in advertising revenue.

Other creators took notice, remarking that his revenue was astronomical compared to theirs. When the top creator is getting 50-250x the revenue share % the average creator is seeing, you know the deck is stacked against you.

Google News, Google Discover, and other news feeds have always been a bit spammy, but they have become utterly worthless in the last 6-9 months, depending on your topic of choice. Most are reddit, stack overflow, twitter, or Tumblr posts, annotated AI-written copy, and loaded down with as many ads as the page can hold while still loading approximately well on mobile. This has turned the casual article or two you read while waiting for a bus/train/bathroom stall at a bar, into a call for eye bleach, as you furiously scroll through a dozen ads just to get to the text the clickbait article promised.

If it was just hiding ads on Twitter, ad-stuffed articles on Google, or goofy art shared to Facebook, this might not be worth writing about. But the scams are already overwhelming Twitter, Amazon, Meta, and Google. The “Kidz Bop” version of thousands of books have popped up on Amazon, fakes written by AI, submitted by famous-sounding authors or outright stolen identities. Amazon is doing nothing, so authors have no recourse when their work used to churn out filth that has half of their name on it.

On Twitter, bots are mass-following each other, replying to each other, pushing sex, Nazi conspiracies, diet pills, and moody gold. Post-IPO Reddit has proclaimed AI to be its path to profitability, which makes all of its human moderators (that weren’t given shares of stock) just ecstatic that they get to be spam filters on top of the internet hall monitor work they’re already doing.

All of this compiles into a world where the hum/noise of social media isn’t the hustle and bustle of a market, the chatter of a bar or busy restaurant, or the hum of a productive shared office space. It’s the uncanny valley of interaction, it is Blade Runners killing Replicants. It is your favorite game critic telling you they LOVE AFK Arena even though they only play Diplomacy & hex-focused war simulators. It is HAL 9000 saying “I’m afraid I can’t do that Dave,” but in this case, D.A.V.E. is just another worker bot on the ship. The Digital Dark Forest has been replaced by the Synthetic Light Forest. The people haven’t returned, but they’ve been replaced by bots instead of the wind, whistling through the hollows.

If you’re sick of Google, here are alternative search engines that caught the eye of a SEO. They might not be any better, just a different flavor of algoweird, depending on what you’re looking for.

• Startpage is the closest thing to a Google-like experience with no search history, no ads, etc. Their tech explanation could be all lies, but they could just hope to build marketshare like Ecosia.

• Yep claims to give 90% of revenue back to creators, which could be huge if they build scale and don’t rip off the community.

• Openverse.org is a particularly useful search engine that serves up just open source, creative commons, and ok to use media for use. No AI material, and no iStock/Getty vampirism!

• You.com combines a search box with a ChatGPT style response. Adding links and media when needed, the responses are helpful, if a bit cluttered/redundant. Perplexity.ai is a similar offering, one where you can see the ads they’re going to put in if they get popular, if you squint just hard enough.

Not sure where we go from here, but pay for human-created things if you can. I don’t know anyone who honestly prefers machine-curated content or AI-created art, but here we are. Instead of complaining about it, I’m going to keep building the alternative, and sharing it (when I get up the courage to) to other humans. That’s all we can do right? There can’t be a better world if we don’t build it. Ad-free, human curated, brought to you by Factual Dispatch, as long as that’s a thing humans still want.

Instead of continuing this diatribe, I’m going to shut up and link “We Want Your Soul” by Adam Freeland, with one of the best monologues from comedian Bill Hicks (minus the r-word).

Keep your head up,tn