Removing Ollama From My Mac

The AI experiment in my "laboratory" is over. I've been running Ollama locally on my laptop, as in:- an 8G, local, offline language model, running on local, offline command line software. Don't get me started about the carbon pollution environmental disaster being created by serverside queries to AI like, "Does my partner really love me." The local-only model seemed like an affordable and accountable approach and my M1 Mac is energy efficient fast enough that even really tricky queries were sorted way under 2 minutes. The thing is, I never got a single, useful response to a query. Not one, that was fit for purpose, anyway, out of Ollama's "mouth."

I figured that between my ability to describe a problem, and its probable solution, architecturally and simply, might result in useful code fragments for my various microcontroller coding and OpenSCAD design tasks. Not one. The work required to make anything useful was at least as much as designing it directly from my own rough scope, if not double in most cases. Then the roughly 2/3 of all attempts it just simply tripped balls, hallucinating like Timothy Leary at a 1968 love-in.

OK, lets try some creative writing, then, "Ollama, outline a novel from (a cat command linked text file of a story arc) and 20 chapter titles." The bloody thing couldn't even stick to the brief! Wobbling off out of the science fiction scope into period Wellsian arcs unrelated even to the story requested.

Where it worked was in really basic general stuff. Essentially working best at the stuff that was general, theoretical and unimaginative. Stuff I already knew, didn't need "help" with. Essentially, it has no creativity, no imagination or synthesis, just a bunch of answers assembled from pre-existing data, in a probablistic way. It guesses answers, essentially. Siri is more intelligent. Siri isn't AI, it's just an advanced, text and speech activated search tool with a massive database.

What prompted me to perform this task was, firstly, today's weekly "Where's Your Ed At" newsletter, about how Glodman Sachs has tipped the AI bubble is about to burst, and 2 Newsletters (last week and this week) from Mary Spender about how Universal Music Group, Sony Music Entertainment, and Warner Music Group are suing the AI music start-ups Suno and Udio for copyright infringements. These 2 stories address the two major problems of AI:- AI is built from using massive amounts of data, including data from the music industry, research prepress archives, the entire internet, essentially, and that AI is increasingly an investment in diminishing returns. The race to the bottom is approaching zero, the money and infrastructure invested is on a trajectory like Apollo 11's in 1969.

AI is a sideshow trick. It's an overblown search engine, stealing the digital rights of anybody who has ever created anything, regurgitating probablistic answers that seem logical enough at first, in an attempt to fleece venture capital, all the while hoping to make creativity something venture capital need never pay for again.

And this is the problem I have with modern capitalism. Money is the only motive, not value, not community benefit, simply cash. The venture capital firms that funded Apple's growth, or that backed Facebook in the early days, or Elon's desire to buy Twitter, were seeking money for the power it brings. The value of the stocks they invested in were seen solely as a way to hoover up money, all at once, from everywhere. All of the money!

Some individual investers might see the value the product creates, but venture capital firms are predators. In small numbers, useful to the ecosystem, to weed out the weak and test the adaptivity of prey species mutations, but really a minor part of an ecosystem that is actually dominated by vegetation and myceleum. The worker species. Capitalism is a system that gives too much weight to the predator (the too rich and too powerful) and the prey (the middle class), while ignoring the needs human needs of all classes, but especially the general needs of the working and itinerant classes. Capitalist wealth does trickle down, but never enough to create security for the very bedrock of community, the majority of people working to create the wealth that gushes up while tiny leaks of overblown largess here and there trickle down.

AI capitalists would make the very researchers in universities, trying to understand what thought and synthisis of ideas actually is, obsolete if they could. Mr Moneybags doesn't want to spend money, he wants to have a machine that creates things that will make him money forever, without having to spend money, forgetting that, if he eliminates skilled people, the machine will not pay for his goods or services. Capitalism is essentially eugenics with money. Instead of eliminating [insert allegedly "undesirables" here], eliminate the poor. Which, in AI world, is even people who create, think, care or make change for the better. Money, becomes the only measure, the desire for monopoly, the only goal. What a lonely world the ultimate capitalist will ultimately live in.

So, if you thought, like me, that trying an AI on your local machine, disconnected from "big AI" was a possibly useful tool but, like me, have come to realise it's yet another race to the bottom of enslaving, then discarding us, here is how you remove Ollama...

• Open a Terminal (Mac only at this stage, 'Nixers probably don't need help with this),
• type "rm /usr/local/bin/ollama" at the command prompt then "Enter",
• type "rm -rf ~/Library/Application\ Support/Ollama" then "Enter",
• then "rm -rf ~/Library/Application\ Support/Ollama" then "Enter",
• finally "rm -rf .Applications/Ollama.app", then "Enter" and Ollama should be gone completely.
• You can test it is gone by trying "ollama" in the prompt, if you get "command not found: ollama" It's gone.

For windows and linux users, Googling about should result in similar methods.

Resist the desire to be "lazy", creativity is meant to be a struggle. It's not a struggle like poverty, it's a struggle with at least some privilege, if not great financial security. The struggle in creativity is one of understanding our goals, our motives, of understanding our humanity and how to express it. Machines that can build our ideas, like 3D printers or musical instruments, chisels, brushes, etc, are marvels, useful tools of expression. Handing of thought, brainstorming or other cognitive processes to a machine is to devalue yourself, to devalue your intelligence, intellectual and emotional.

Machine-conceived art is not art, at least not human art. It can probably be interesting or challenging but it has no root in life, and art is living minds at play - both serious and whimsical. A machine cannot currently play, or think - it can run or stop, give an answer that is right or wrong in varying probability but it cannot synthesise something new and unique without a pre-existing formula for type. Maybe one day, a general artificial intelligence will have the ability to free run in thought, result in a synthesised concept that is expressive of its very existance. A haiku, perhaps, that captures something a human could not understand about the machine's experience, without the machine first describing it. Then, we will be in a first contact situation with a new species. That machine will be alive. It will not be like us, yet it will. I will treat that machine as a living thinking being until I can prove it not such. Right now, a "thinking computer" is as elusive as it ever has been. ChatGPT and the others are little more than Eliza running on hardware that is a solution akin to opening a peanut with a pile driver when one's own fingers would suffice. I'm not the first to make this comparison.

That is all.