If that's true, we're now unavoidably heading for a full-on AI wipeout, in which there is no incentive for humans to provide any non-commercial information or valuable imagery, and everything henceforth is just a rehash of pre-existing material.
As the Disgruntled Souls of Twitter were stressing enthusiastically over which scroll-zombified, digital-credit-scoring, transactional labour-mill they'd allow to waste their lives next, a neighbouring Silicon Villain was releasing a new toy. And oh, did the public want to play with it!
The product? ChatGPT. A chatbot whose friendly verbal spewings have already become internet-famous.
Actually, this flagship release from OpenAI is considerably more than a toy. Conceptually, the supercharged chatbot is really just the prodigy lovechild of Wolfram Alpha and Mitsuku. But it's phenomenally sophisticated, and the number of people its legacy could render workless is truly frightening.
Whilst ChatGPT is pitched simply as a human-like query and response system, the legacy of its far-reaching capabilities threatens publishers, creatives, information sources, advisors, forums, social networks, search engines... Yes, search engines.
True, ChatGPT has been engineered to avoid treading on some of the key commercial roles for websearch. And how nice of the makers to consider Google's ass whilst chucking the rest of humanity under a bus, might I say. But that brotherly love for Silicon Valley neighbours will only apply in the now. The long-term legacy of this technology is lethal for websearch as we know it.
Compared with a search engine, ChatGPT offers frictionless evasion of link-clicks. Hugely reduced trial and error. Elimination of tiresome, off-point SEO-stuffing, etc. ChatGPT responds only to the demand. It doesn't meander or ramble. Doesn't (yet) try to send you to Amazon or serve you a newsletter popup. It gets straight to the point, and it delivers no superfluous bulk. These benefits will, in time, convenience-whip the public into ditching Google's current high-friction system for most forms of discovery.
And at present, at least, the insultingly poor quality of websearch results is not particularly helping Google's cause.
But the sunset of 1990s-style discovery will not happen overnight. For a start, ChatGPT is set to run as a paid service, and even during its current free trial period it requires the user to create an account and log in. That's friction in itself. And more profoundly, the back end is much more work-intensive than a typical search engine - to the point where completely replacing Google would at the present time be impractical.
AI IS SYSTEMIC THEFT. And in a world of systemic theft, all the valuables eventually get locked away.
But we're already seeing that statistically, people overwhelmingly prefer completely incorrect but plausible responses from ChatGPT over and above valid information via Google. That indicates very considerable demand for frictionless query and response. And demand is a powerful force indeed.
It's not just the natural-language input and lack of in-process friction that people prefer either. ChatGPT's presentation is, particularly in mathematically-driven subjects, vastly more digestible than the kind of presentation typically seen from mathematically- or scientifically-minded people. ChatGPT will not only explain a computer programming task more clearly than the average programmer - it will actually write the code for you.
And that's not all it will write for you. It'll write an advert, a love letter, a newspaper article, an assignment essay, a job application... Whilst I find it quite amusing that job applicants are about to start chatbotting employers' chatbots, and that the entire OnlyFans messaging system will soon comprise nil but text-generators having conversations with themselves, the implications of all this are unthinkably dystopian. Even setting aside the jobs that will inevitably be lost, the idea of a world in which most future reading matter will be written by a robot, is sobering in the extreme. Even novels will probably be written by bots before the middle of the century.
BEYOND THE NOVELTY...
We should, however, keep the current state of play in perspective. It's not 2050 - it's 2022, and the suppressed, part-time Buddhist in me says we should concentrate on the present, and make what we can of it.
If ChatGPT were a human being, it would be referenced along the lines of "noob" or "blagger". So why are people showering it in accolades? Because there's novelty in seeing a machine convincingly mimic human communication. Especially when it succinctly storifies. ChatGPT achieves that realistic feel using a combination of straightforward parroting, linguistic substitution, and prediction. The scope to poll an ultra-large language model enables the routine to attain consensus if not truth, and undeniably, the results can look mindblowingly intelligent. Especially prefaced by a user's query or command in natural human form.
So ChatGPT's first response tends to floor people. But its hundredth? Not so much. It'll probably be some time before tools of this type can impress beyond the novelty phase. The style of ChatGPT soon becomes recognisable, and the software is very easy to trick into stupid responses. Oh, and the bot will need a trifle more than capital investment when it comes to creative tasks.
Its poetry is embarrassing, and if you take it down the path of jokes it either writes tumbleweed or flat-out plagiarises. Mostly the latter. In fact, even if you ask it to write an original joke it will plagiarise - potentially misleading the user into copyright breach. It heavily repeats the same loop of plagiarised jokes, too. If the human version of ChatGPT managed to make it through one party, they definitely wouldn't be getting an invite to the next.
Because ChatGPT is not human, it has no instinct. No sense of style. It can't assess whether a line has punch. Whether a statement is funny. It can't imagine. Only regurgitate, reword and algorithmically predict.
Ah yes, the 'benefits' of AI. Lost livelihoods, more surveillance, more subjugation, more poverty, more misery, more prejudice, more exploitation, more anger, more concentration of wealth, nosediving quality, and a plunge in society's mental health. What more could a bunch of late-stage capitalists want?
In the short term, this could actually help good creative writers. If most bad to mediocre writers are using chatbots to generate text in a style that's recognisable, boring and repetitive, then impassioned, inventive, artistic and funny writers are likely to stand out more. My suspicion is that a combination of laziness and delusion re ChatGPT's capabilities will drive many more producers of written content to shift towards automation in the next year or so. Especially a wider section of content-marketers on limited budgets.
The good news for pro-humans here in the present, is that whilst bot-generated content appears acceptable in isolation, it's essentially unshareable, and that severely limits its success potential. The criteria for sharing are quite different from the criteria for technical competence. People don't share content because the grammar is perfect, or because the body text "does what it says on the tin". They share it because it connects with their emotions or awakens them in some way. Maybe it's nostalgic, or relatable, or just funny. These are very human properties, and bots don't understand them at all.
One of the things I've noticed in recent months is that some of my old articles on the music site Planet Botch are gaining new shares. Around the start of the 2020s I was seriously considering rewriting some of these throwbacks from the early 2010s. Not because the information is out of date - it isn't. But because I'd come to feel they were too personally-biased. Most of the articles seemed to have died, and I thought people wanted a more encyclopedic feel, because that's the kind of content Google was ranking. But it now looks like people are so cheesed off with mechanical copy, that they prefer less well-written articles with some human perspective.
The rise of AI actually looks to have increased the sharing of 'unencyclopedic' articles, and that's positive news for the immediate term.
THE LONGER TERM...
But looking further ahead, the prognosis is dire. Quote unquote, AI IS SYSTEMIC THEFT. And in a world of systemic theft, all the valuables eventually get locked away.
ChatGPT steals some content almost verbatim, and no intelligent creator is under any illusions about how these tools work. Even if they're not recognisably stealing, AI bots are only thinly disguising real people's statements with linguistic substitution. It's just a simple translation process. Except the tongue doesn't necessarily change. Just as these bots can translate English into, say, Spanish, they can translate English into a different version of English. So what looks like original output is really just an algorithmic spin.
A creative backlash is already in motion, and in fact, it would not be over-dramatising to say that this could be the beginning of the end for openly accessible, human-made online content.
Artists have withdrawn an immeasurable volume of work from the open web in the wake of tools like Stable Diffusion - the AI art "generator". They're gating off entire archives to avoid having their work stolen and rehashed. Not that this will protect them from AI scraping. They'll soon learn that when you gate an archive with value to AI companies, the AI companies will just go through the back door and plunder the content directly via the storage and delivery chain.
But we can see the sentiment. And once everyone is clear that the only way to stop tech monsters from sucking work into an anti-human AI project is not to upload it at all, that is what will happen. People will not upload at all. Why would they? What's the point in being an unpaid servant to a piece of software which ends up getting all the credit, applause and financial reward?
STALEMATE OF DOOM
It's here that we revisit the threat to Google and other 'conventional' search engines. The future of such utilities looks bleak, and you don't have to reach very far into that future before an impending stalemate rears its head.
We wouldn't expect Google to stand by whilst the legacy of ChatGPT gobbles up its search traffic. But we know that if Google Search doesn't match the convenience of ChatGPT, it cannot survive in the long term. Google is also now facing a massive threat from Twitter, as the Bird Site explores ways to improve the user-friendliness of its search engine.
People are generally unaware of Twitter Search's power. But it's much better than Google, and that will quickly dawn on the world if the promises of Elon Musk come to fruition and the Twitter Search interface is radically improved, infused with AI, etc. Google Search will have to act in order to compete. And however much you may dislike Musk, he's not some EFF-worshipping sap who will lay off Google because the lobbying collective advises him to. If he thinks he can kick Google's ass he will not hold back.
The problem is, a shift from Google to match ChatGPT's convenience would result in websearch engines that send no traffic to sources. So er, what percentage of publishers can you imagine creating for search engines that send them no traffic, and which merely feed their work into a bot that seeks to render their output redundant? I'm thinking close to nil. Online content production as we know it would be over.
The scary part is that it now appears Google will be left with no option but to migrate to a zero-referral model across much of its territory. If that's true, we're now unavoidably heading for a full-on AI wipeout, in which there is no incentive for humans to provide any non-commercial information or valuable imagery, and everything henceforth is just a rehash of pre-existing material.
Stalemate.
Sources won't move because they know AI will steal their work. AI can't move because it has no fresh sources. If you think Silicon Valley's anti-creator culture has decimated the tide of valuable content, wait until you see the barren void that late-stage anti-humanism is about to usher in.
Enjoy the now, for tomorrow, the slow-mo car crash of artificial intelligence begins advancing through its initial frames. An ethical disaster we must all watch, knowing there isn't a thing we can say or do to prevent it.