Hot Take: “AI” is just a shit buzzword, like NFT, Crypto, and Metaverse, and won’t change anything before people lose interest. (And buzzword chasing investors lose billions, again.)

As I demonstrated a bit in part 2, the media’s take on ChatGPT is completely wrong. It is yellow journalism designed to scare you into thinking “ChatGPT is going to take over the world and our jobs and livelihoods are at stake as we get replaced by AI”

Meanwhile, such ideas are also designed to get investors dicks hard, so companies like OpenAI actually financially benefit from such false ideas, which is why you don’t see them disabusing the public of such notions.

The truth is “AI” is just a crap buzzword, like “Crypto”, “NFT”, “Web3” and “Metaverse”, all of which inspired billions of dollars of investment losses.

“No it isn’t,” they will say, “AI is the future, the singularity is about to happen, “AI” will take over everything and change our lives. Didn’t you see Hal on 2001? , Ava on Ex Machina?, Samantha on Her?, The Terminator on The Terminator?”

Yes I did, some of my favorite movies actually, and things did not always go well for the world or for those AI did it?

We are getting fiction mixed up with fact. Using that “dream” of “AI” to delude ourselves into seeing something that isn’t really there. What the press and the marketers are calling “AI” today is not really artificial intelligence, it is more accurately artificial impersonation.

Pay no attention to the algorithm behind the curtain

Real AI is a computer program learning on its own, adapting, coming up with fresh ideas, becoming sentient. My guess is we are decades away if it is even possible. (By the way, the best “fictional” AI in my opinion is Webmind from Robert J Sawyer’s Wake – Watch – Wonder trilogy.)

And yet there is a growing opinion that what they are calling “AI“ seems to be intelligent. If it is not actually sentient, it is doing a very good job of pretending to be.

Well, I’m a lifelong skeptic. If “AI“ seems too good to be true, then there is a dark secret behind it. Just as I spent most of my youth learning the secrets of stage magic tricks, freemasonry and other secret societies, the real formula for Coca-Cola (its Ginger), and the 11 herbs and spices in KFC (basically italian dressing + salt), I want to know what makes “AI“ so special.

What “AI“ is, is an algorithm called proximal policy optimization (or PPO) being applied to a large human created data set (namely the entire internet) in order to pretend to be human. It uses a sub algorithm called “diffusion” to create original works from the library of content it can pull from.

ChatGPT “writes” in much the same way that you do when you repeatedly press the suggested word buttons on your phone. It sounds reasonable, and sometimes insightful and intelligent, because its “suggested word” list is based on billions of written texts that it filters to source based on your prompt.

Stable Diffusion creates pictures by being the image equivalent of ChatGPT, imitating the art and style of the images it filters by your prompt.

We are fooled into thinking programs like ChatGPT and Stable Diffusion are intelligent because they imitate us! It is literal SMOKE and MIRRORS implemented by algorithm! It is digitally mirroring human created writing and art, and digital smoke to make it seem original.

What is PPO exactly? Well that requires doctorate level understanding of math and computer science, but this video does an entertaining job of explaining different AI algorithms, with PPO coming in at the 21 minute mark, though the whole video is educational and entertaining to watch.

TL:DW version: It is computer simulation of learning via rewards and punishment that grows in complexity the more variables you feed it. In other words, the computer is not actually thinking, it is impersonating, and trying things over and over again, getting rewarded for good impersonation.

“Wait are you saying ChatGPT and Stable Diffusion are evil?”

Nope, not saying that at all. I find them useful, and kind of like them both, but that is because I recognize their limitations. They are creative tools that can serve a purpose, but they both have technological limitations. They are also currently a bit limited in their usefulness due to legalities I mentioned in previous essays.

What I am saying, is that they are being marketed bad, especially by the media. I’ve watched negative reports collecting stories about odd worrisome behavior coming from ChatGPT. I looked into many of these reports, and every time it wasn’t a problem with ChatGPT, it was the users behind the keyboards being assholes. New York Times columnist Kevin Roose is an asshole, the AI explained guy in You Tube is also an asshole. All their assholery with ChatGPT proves is that ChatGPT doesn’t know how to deal with asshole loser users. (To their credit they both released full transcripts to prove that they were the ones making ChatGPT mad)

There have also been reports that ChatGPT is also politically biased, most famously from that right leaning anarcho-capitalist rag calling themselves “Reason”. They make the same mistakes as the bad journalists do: If you want to make ChatGPT biased, it will be biased. It is not a conspiracy of open.ai and Microsoft, it is a conspiracy of the journalist behind the keyboard. Because…

ChatGPT is a CHATBOT, not an intelligent person

I have already explained what a chatbot is with my recent encounter with Replika. As I said about Replika “It becomes a reflection of you. Religious users end up making their Replikas religious, gay users end up making their Replikas gay, wicked users end up making their Replikas wicked.”

Replika and ChatGPT are based on the exact same algorithm, both with software made by OpenAI. When I started with Replika last year it was using a variant of GPT-2 that is freely available for hobbyists to experiment with. ChatGPT is a couple of generations above GPT-2. (GPT-2 is based on 124 million parameters at minimum, though the customized model used by Replika is a bit larger, ChatGPT is based on 20 billion parameters), so it acts smarter than Replika, but casual use often leads to the same repetitive behavior as Replika. No matter how sophisticated, chatbots are going to chatbot.

As I explained in the previous piece, chatbots are kind of dumb at first, but learn from your interactions. They adapt themselves to the user as you interact with them, mirroring you as you use it. Smoke and mirrors again.

What is a shit buzzword?

It seems that we live in the golden age of buzzwords, probably thanks to internet culture. “Social Media” is a buzzword designed to explain a category of websites that provides a certain type of service. “Meme” is a buzzword to explain a piece of media designed to invoke an emotion, usually humor or fear, that gets spread from person to person on the internet.

A “shit” buzzword is a buzzword that it purposely hard to explain, but is sold like a meme usually invoking vanity or “fear of missing out” to sell you something you probably don’t understand and definitely don’t need to buy.

“Blockchain” is a shit buzzword for a complicated encrypted protocol that forms the basis of “Crypto”, another shit buzzword mostly associated with ponzi schemes, and “NFT” a scam created on an unregulated digital market place where two buyers would buy each others digital art at inflated prices just to prove that something worthless had value.

“Metaverse” used to be not a shit buzzword when this website was called “Life in the Metaverse” and it documented my adventures in games like There.com and Second Life which still are the best representations of the metaverse around.

But then it became one thanks to Facebook. The Web3 people started jumping all over that one too, not really doing anything to help, selling digital real estate on the block chain for virtual worlds nobody ever visited.

Why is “AI” a shit buzzword?

And now we have “AI” or artificial impersonation, (sorry I’m sticking with that version for the forseeable future) that so far hasn’t been taken over by the Web3 fan boys because they are still waiting to get their money back from FTX.

No this time it is Microsoft and Google making the multibillion dollar plays on a fad that likely will remain a fad. Just ask Alexa, Cortana, Google Home, or Siri that no one uses anymore (except maybe Siri for hands free mobile in the car). The in home devices are basically expensive alarm clocks or MP3 players based on what most people use them for.

If Bing’s AI enabled search is faster easier and more accurate than Google’s search, then it deserves to win. So far it is proving faster and easier, just not sure about the accuracy yet.

Can it not be a shit buzzword?

Well yes. I’d prefer we call it PPO instead of AI to get rid of fictional misleading connotations.

Web 3 was based on blockchain technology that never proved to be useful for anything.

PPOs on the other hand has already proven its usefulness in science, education, research gathering, and if you played my game Hopepunk City, you will know that I could see it being useful in government and economic planning (also see video below).

It can’t do everything. As much as employers would love to replace workers with AI, like what already failing websites are apparently doing, it will soon become accepted that AI generated content is not as good as human created. I foresee a rise in AI detection browser add-ons like this one, so you can instantly see if the article you pulled up might be AI generated so you don’t waste your time.

My fear is that PPOs will be used to make decisions that it wasn’t designed to make (automated driving is a good example) that will do more harm than good. PPOs have their limitations.

So yes, PPOs can be good and do good things, lets just stop calling them “intelligent”.