Stop The Slop

AI seems to be at an all-time boiling point, which to me is kind of insane. No other technology before this has caused such intense societal emotion, albeit with downright anger. Even a company like Meta, which could easily eclipse AI in terms of controversies and societal damage (as far as I know), would seem to be almost popular compared to AI.

But the problem with AI first and foremost for me is the slop. Even Dario Amodei the CEO of Anthropic refers to slop in his recent Bloomberg interview. All this video, image, audio junk that “AI” can produce ruins the product that is AI. The killer “feature” of AI is synthesis and execution of computer systems and data. Which is the most boring thing you could possibly say. But it’s true. Projects like AlphaFold and tools like Claude Code are what make me excited about AI. It’s giving smart people tools to create amazing things. Giving a coder Claude Code allows them to orchestrate and create things at scale that would take a team. Things like AlphaFold are able to help scientists make breakthroughs and advance modern medicine.

But give someone who just wants to make AdSense YouTube money Sora we get fruit Love Island. Since these generative AI videos and images take so much compute, it just makes the datacenter and memory crisis even more infuriating knowing this is a cause.

This is honestly why I think OpenAI is floundering where Anthropic seems to be thriving. Anthropic has never had any video or image generation; it has been all about boring tools. But people swear by these boring tools. Whereas OpenAI makes these cool flashy products that seem to just burn money without establishing more market share. Whereas Google DeepMind has the infinite money reserves to explore things like Nano Banana and Gemini Omni, they are still making big strides in the boring stuff Anthropic has focused on. To them, it feels like more of a way of just keeping pace in case this AI slop stuff does take off in a meaningful way.

Back to my earlier example though, AI is an amazing tool in the hands of people who know how to create amazing things in their domain. But an artist doesn’t just pay someone else to paint for them. These AI tools can have a place in creative applications, but it would be in boring stuff like masking in video editing, background removal in photo editing, and other very tedious tasks that have nothing to do with the actual process of creation, but more just about speeding up the menial.

Programmers can still code without Claude; how much they use it is up to them. Human in the loop is still incredibly important for all these things.

Also, just because someone might not be a senior dev or scientist doesn’t mean they can’t use AI. In fact, it can really help close the skill gap, but people wanting to achieve a higher level of work will create things that matter with AI. Whereas what we see now is people just making things because it costs a monthly subscription and about a paragraph of text to make money with it—that is an ugly sore that the industry needs to start treating.

This of course isn’t the only thing wrong with AI, but it’s something that, when I think of AI, the slop is what makes me feel gross in even saying anything positive about it.