How long have we got before the agent output is enshittified?
nothing_is_real
15 points
16 comments
November 17, 2026
Related Discussions
Found 5 related stories in 292.0ms across 13,334 title embeddings via pgvector HNSW
- Progressive disclosure in agents to make them 2.3x faster suchintan · 15 pts · October 24, 2026 · 51% similar
- How version control will evolve for the agent boom tapanjk · 46 pts · July 09, 2026 · 51% similar
- AI Agents and the Refactoring That Never Happens rosenfeld · 42 pts · September 02, 2026 · 49% similar
- Here be monsters: What do you still need to know when agents do the work? adrq · 166 pts · October 27, 2026 · 49% similar
- Why are coding agents so dumb? mtlynch · 84 pts · October 09, 2026 · 49% similar
Discussion Highlights (7 comments)
boothby
Hopefully it happens before neural interfaces take off so people see the light. No, I jest, people don’t learn. The neural interface will be the unshittified option for a time and people will flock to it
IggleSniggle
It has already happened, just not in the way you are imagining here, imho.
vivzkestrel
- stupid question - does every piece of tech every conceived go through this cycle of enshittification? - has anyone actually looked into all the things on the tech side that have released since 1990 and calculated how many of them got enshittified?
sidsud
this could intersect with model collapse[1] - when we run out of human-generated training data [1] https://en.wikipedia.org/wiki/Model_collapse
officeplant
Two years ago? :3
gosub100
I think the next step is a form of "drink verification can to continue": A map is an associative container typically implemented using a set. If you'd like to see an example in C++, please answer 2 of the following questions: Then proceed to build on existing profiles of people (of course while selling them to third-parties) and their behaviors. Or put them to work on heuristics and learning that they don't have enough training data on or contemporary things that haven't been ingested to the data center yet. This will be annoying just like YouTube ads, so they will sell you the solution in the form of a higher priced tier. That is how I predict AI will become enshittified.
mft_
Claude has twice this week refused to give me a summary about particular cancer types, and the drugs used to treat them. This would all be based on public domain information, and is far from anything remotely edgy or controversial. Add to this the reported refusals to engage in activities related to reverse engineering, and I think it’s safe to say that Claude’s enshittification is already well under way. I’m already wondering which service, harness and models to use as a replacement.