AI
A comeback blog
This is not a thinkpiece about AI and society. This is not railing against AI doing the creative work and potentially making artists of various kinds obsolete. This is not about how humans are uniquely positioned to produce better creative output than AI because of some life force.
This is about how the rising popularity of AI feels a lot similar to the early days of the internet, and what we can learn about it.
Remember the old internet? Not the corporate wasteland it seems today. I'm talking about game forums and IRC chat rooms. About music torrenting and how it routinely gave you viruses (virii? lol). About MySpace which came around a bit later. An internet which was difficult getting around unless you were savvy enough. When gatekeeping meant keeping the n00bs and tourists away from your interests in a more real sense.
More hardcore users might remember Eternal September but I wasn't born yet. My "Eternal September", and really my internet journey, began with Facebook. Regular people now had a reason to go on the internet which didn't involve computer and internet knowledge. No, Facebook succeeded in giving a lot of people a reason to be online - to communicate with their friends.
ChatGPT was launched in November 2022. We've had an Eternal November of AI since.
AI research has actually existed since the 60s. Much of early work was about mimicking human behavior in response to stimuli. An introductory AI class explaining the origin of AI feels eerily similar to an introductory psychology class. Source: me, who took an AI class at 2 and Intro Psych at 3. Skipping to the mid 2010s, Machine Learning is the in-thing, Data Scientist is a viable career path. Colleges are abuzz with RNNs, LSTMs and some prototype Transformers stuff too. Every fresher has a "Detecting Cancer in Breast Tissue" project lifted from someone on GitHub.
AI wasn't for the normies yet.
After a brief collective amnesia of CS students learning blockchain and crypto, AI had its Facebook moment. ChatGPT showed that regular people had a use for AI: a thousand monkeys that can write Hamlet on tap. And also a sappy romance, a racist rant, an emotional apology, an SEO-optimized blog post. With minimal effort.
The AI arms race has been quite similar to the corporate social media arms race. Different platforms started emerging - Instagram/Stable Diffusion, Twitter/Character.ai. Slowly, they started becoming less distinct from each other. Acquisitions became the name of the game. Even Cisco and NVIDIA's respective meteoric rises have been similar, although Cisco's rise and fall happened slightly before the social media boom.
And just like corporate social media, AI too has a counter-revolutionary movement.
Just like computer nerds huddling on the "fediverse" (don't look into it, it's nothing cool I promise), AI nerds huddle on forums like r/LocalLLaMA. Their purpose - to hold the power of the AI in the palm of their GPUs. The ideology feels similar too - corporate controls too much of the software and resources, which it can restrict at any time. So we must make running AI models locally better with the power of the open source. Not raining on their parade, but unless this paper lets us coordinate training a GPT-4 sized model across the world, we won't achieve true independence from corporate AI.
So, what do you, the reader, do with this information? We might soon arrive at a inflection point when we achieve AGI - a kind of intelligence quite like a human's. It might destablize society or simply fade into the past with more promising technology come along. But it might impact your life and your source of income. Be AI-literate, ride the wave, learn to use the technology to your advantage. We are already in a time where basic AI prompting skills are becoming important to do your routine tasks and homework. It might take you some work, but it will be worth it. Smelling AI-generated content will allow you to be rational in your everyday decisions.
For the software engineers: I implore you to read this blog. Technological fads follow a cycle; our job as software engineers is to build resilient and scalable software. AI might keep you relevant now, but something more shiny will come along in the future. Your core skills should be your focus of improvement, AI/blockchain/zero-knowledge/quantum are simply the friends we make along the way.