From Newsgroup: comp.misc
From the -2from irritating to unusable-+ department:
Feed: Luke Smith's Webpage on Luke Smith
Title: Disenchantment With the Post-AI Internet
Date: Fri, 21 Aug 2026 19:45:09 +0000
Link:
https://lukesmith.xyz/articles/disenchantment-with-the-post-ai-internet/
The internet has gone from annoying to unusable. It once could be
insipid or dangerously temptaciousrCoNow I canrCOt even bear to be on it.
One would be justified in wondering why I havenrCOt posted videos or
articles in a while. IrCOve taken breaks in the pastrCosometimes for months, but starting around 2023 I nearly entirely went radio silent. After a
bit of a personal break, the internet I came back to was entirely
different. The bad seeds were already there when I left, but they had
come to blossom in quite ugly fruit.
I had complained in 2021[1] that search engines had already become
entirely useless due to algorithmically-generated sites that even back
then were probably most of the internet. Even then, searching nearly any
query would be met with a long string of fake sites with algorithmically generated text to garner clicks and ads.
Now LLMs/AI have so significantly exacerbated this ratio of signal to
noise that there really are no rCLresultsrCY in search engines. Search
engine results are simply 100% auto-generated. Nearly every page you see
in the results of a very specific query might have never been seen by
another human.
The Narrative
One of the most dangerous social, political, academic, scientific,
religious, spiritual, economic etc. outcomes of this is that LLMs/AI
have now taken a snapshot of the standard political and social
narratives at the instant of their inception and that will serve as the rCLinputrCY and the proving ground of narrative norms of the future. When everyonerCOs favorite AIs are all trained on the same dataset and thus confabulate the consensus views of that data set (plus or minus the
biases of the wielder), this will calcify a certain set of rCLright
answersrCY or rCLacademic/social/etc. consensus.rCY
This is the dangerous thing that Wikipedia has already done in part, but instead of being on one particular site that any informed person knows
should not be trusted on certain issues, it will be the background
narrative for the entire internet and all people who rely on it for
their thoughtless decisions.
Some of my critiques of the way that academia or media have worked in
the past are now being absolutely dwarfed by the even greater problem
that there will now be an AI-enforced social norm of everything from
narratives of history to medicine to economics. There will be variation,
but what is important is that the whole system is incestuous: todayrCOs AI output not only rCLeducatesrCY a new generation of rCLthinkers,rCY but also directly serves as the input to new AI algorithms.
This ultimately means convergence on one set narrative, one official rCLWikipedianrCY narrative which of course is the product of lying text and word generators that are written and trained to be well calibrated to
rCLsound rightrCY to humans.
Increased, Not Decreased Reliance on Noise
People also use AI more explicitly as a consultant than they ever did
search engines. IrCOve run into a lot of people who troubleshoot very
basic problemsrConor do I mean only technology problems, but other things
as wellrCothrough AI. If anything, the algorithmically-derived nature of
AI and search engines radically decreases the signal to noise ratio in
digital technology, but the illusion of the customized, specific and linguistically competent and confident AIs goes very far to make people
believe in the magic trick.
So at the moment where our technology is least helpful and most empty or random, we are trusting it ever more.
Even YouTube
When I did the video above, I believe I noted that YouTube (or any video
site) is at least a little more shielded from algorithmic content than
normal text websites. After all, it even now is still much easier to auto-generate text websites than videos, and although there were some algorithmically generated YouTube channels around even then, LLMs and
other tools have made fake videos now the norm.
This difference has been very stark to me in my hiatus from the
internet. Suddenly, in 2025, when I would be watching a very specific
video on a rare and arcane topic, I would see recommended videos that at
least by title and thumbnail were precisely what IrCOd like to seerCoand not little 3 minute Wikipedia readsrColong videos that certainly looked to be
what I should want.
Invariably, they were AI. AI vibe-coders have created entire fake people
and fake personas for algorithmic channels.
The other day, quite ironically, in searching for home-management recommendations, I stumbled across an rCLAmish manrCOsrCY YouTube channel, started recently and with many views. The channelrCOs videorCOs scripts were very interest-piquing and I certainly would have wanted to watch more if
it wasnrCOt obvious that the clips that made up the video, including the rCLAmish manrCY himself were AI-generatedrCothat and the recommendations of
the channel were only either provocative or inert, like all AI.
What is now disturbing is that that one fake rCLAmishrCY channel is the
norm. Every new YouTube channel should be assumed to be fake, because
frankly nearly all of them are. They produce huge numbers of
high-impact, well-targeted videos very quickly.
The People Are Becoming AI.
My relationship with AI is entirely non-consensual. I donrCOt want it. I
never wanted it. It causes nothing but harm.
But what is most perplexing, but really embarrassing to me is that there
are people who in previous times I might have had some degree of trust
in who have enjoyed rCLriding the tigerrCY of AI.
I must report that I have lamentably found out that a number of people
whose work I used to admire have now taken to leaning on AI to do most
of their thinking for them or writing articles of even books for them.
I remember one of the old 2018-ish Twitter accounts I used to like,
Faceberg, making some quite accurate doom prediction: that influencers,
even rCLbasedrCY ones, would be slowly replaced by algorithms that slowly
shift their narrative and shepherd their followers into the control they imagined they would escape. (If someone has the original post, please
send me it.)
As far as IrCOm concerned, the only inaccuracy of this is that the rCLbasedrCY influencers are often doing it to themselves. First, they export their
work and thinking to an AI, but in the same way that someone who used
Google Maps ceases to be able to navigate without their surrogate brain,
the thinker gradually becomes more AI-reliantrCotheir narratives become adulterated without their knowing, or first in a way that doesnrCOt seem
to bias them in another direction. But when the rCLGoogle MapsrCY is taken away, what happens?
The AI hoax is the new COVID hoax. The only silver lining is that you
can easily distinguish who you can trust and who you canrCOt based on how
they react to it.
Let my reaction be known: I want no part of it. I will never use AI. I
will never lean on it. I will never even consult it. I will never
troubleshoot software with it. I will never use it to condense
information. I will never use it to even generate images or other grunt
work. IrCOm sorry, I can deal with not having it. ItrCOs easy.
I consider AI tainted and every thought filtered through it as tainted.
Code made by it is tainted. Let this be my rCLirrationalrCY abstinence from
AI.
Last year, when a made a video talking about the AI craze, I flippantly referred to it as rCLdemonicrCY with a somewhat jocular smirk. I only take
back the jocular smirk. I will abstain from AI as if it were morally
tainted or demonic.
And, you know I hate to make one of these asides despite the many
atheists who read and watch me, but please do me the pleasure and ask
yourself: Suppose you were an evil incorporeal being that had some
ability to influence matter, but wanted to conceal yourselfrCowhat would
be the best way to influence people without being detected as breaking
the normal chain of physical causality? Certainly the ideal would be a
kind of technology which is nearly entirely unauditable (not like a
typical computer program), but nonetheless capable of imposing complex
ideas and thoughts to humans. LLMs/AI are the first and best example of
such technology.
Social Media Sites
The last pretended bastion of humanity on the internet is the worst part
of it: Social media sites. Social media sites have at least peripheral
interest in keeping some pretense that the people on it are mostly
thinking humans.
Of course, rCLthe algorithm,rCY which is truly the filter through which a person sees the site and the world is entirely automated, uncontrollable
and nowadays based on AI.
But even more I will just state that I feel after that hiatus from the internet, the little interaction I occasionally have with a social media
site seem quite rCLdifferent,rCY uncanny and algorithmic. ItrCOs actually exactly the same as the difference between pre-AI YouTube and post-AI
YouTube.
That is, I might be directed to a thread on Twitter or R*ddit, but
nearly everything seems fake. The humor is different, the target of the
posts unclear, the narratives are a little off, conciseness and wit,
funny or not, are entirely gone.
Of course, in a world when creating a hundred algorithmic accounts is
easier than making two run by humans, this is no surprise. I have no
doubt that we are indeed operating on a rCLdead internet.rCY As I said in a video last year, rCLthe algorithmrCY of these sites functions as a SkinnerrCOs Box which also trains certain behavior even in the minority of accounts genuinely operated by humans.
AI still isnrCOt special.
I do want to be clear, LLMs/AI, despite what their venture capitalists
and cultists would have you believe, are still not transformative. As
specific technologies, in fact, they havenrCOt in the least even changed
the orientation or the speed of the demise of the internet.
LLMs are doing faster what other simpler algorithms were already doing
when I complained in the video above four years ago. I suspect if the
specific technique of LLMs had not been developed, we would have other
creative algorithms to generate the mindless filler on the internet that
is most easily monetizable.
I feel like the cult around AI, and it is very much a cult, is a kind of religious experiment to which attempts to offset the stultification and degradation AI has caused in the internet with some vague optimistic
belief that it is something wonderful or futuristic.
The Sinking Ship of the Internet
Obviously one of the recurrent themes of my online presence is the
important of life, which is the thing that takes place when neither at a computer, nor with an eye on a phone screen.
I almost wish my personal struggle against the internet had lasted
longer: I wish there were still something addictive on it to tempt me. I
wish there would be some goodness in it for goodnessrCOs sake.
There have been times where the internet has been a kind of guilty
pleasure to me that IrCOd have to avoid, but that has entirely passed; it genuinely pains me to be on it. ThererCOs just nothing on it anymore.
I did publicly pine for the return of the 1990rCOs internet, and since
most of the web is now just inert, the only way to actually use it would
be individually browsing individual peoplerCOs sites whose humanity (and abstinence from AI) is beyond question. We effectively have been forced
into that position, but we still need to reinvent discoverability and networking between small sites.
The Forced Pivot to Reality
I used to occasionally have great ideas for things I wanted to do on or
via the internet. As this terrible era has come, however,rCoand perhaps it
is for the bestrCoI think nearly exclusively in terms of what can be done
in reality.
I would rather meet and arrange events with people in real life, to meet
and talk and exchange ideas in local ventures, schools or churches than
I would do the same on a rCLbigger scalerCY on the internet. I will not say that the rCLbig scalerCY of the internet is a bad or non-existent thing, but
I am merely personally reporting that my interest in the
YouTube/internet grind is quite low as a means of accomplishing things.
It still may be the easiest way to reach a large audience, but if that
cannot translate into something real and significant in reality due to
the distraction and noise of the medium of the internet, that might not
means as much as one might think.
So perhaps thatrCOs me saying IrCOm more open to doing live events and meet-ups? ThatrCOs not how I expected this article would end, but I
suppose it is the case and the next logical step. Contact me if
interested and serious.
Links:
[1]:
https://www.youtube.com/watch?v=N8P6MTOQlyk (link)
--- Synchronet 3.22a-Linux NewsLink 1.2