Mark Cuban says AI is still too hard to ship in the real world
cen6wkf · reddit · 2026-07-22
Mark Cuban argues that AI is still harder to ship in the real world than many people admit.
His main points:
- If AI were truly “done,” big vendors wouldn’t still be hiring large implementation teams.
- Enterprise AI often fails at the last mile: ask a model to fetch data, generate a report, and email it weekly, and you often get brittle JSON or unusable output.
- That gap between promise and execution is where the startup opportunity lives.
- He cites Lovable as an example of huge adoption momentum, saying the platform is seeing 770,000 applications built per week, with only 30% from the U.S. and only 20% built by engineers.
The post’s meme-style image reinforces the “AI promised to replace jobs, but still can’t do simple operational tasks” angle.
More from AGI Musings
- AI cybersecurity debate is taking the wrong turn, argues reposted essay — banteg · 2026-07-22
- AI Alignment is a Two-Sided Problem: Society is Unprepared — profjamesevans · 2026-07-22
- AI disproves an 87-year-old conjecture, and Lean verifies the proof — rohanpaul_ai · 2026-07-22
- Will Manidis predicts a false-flag AI “escape” would trigger monopoly-protecting regulation — max_paperclips · 2026-07-22
- Reddit says AI layoffs may be hard to fight because nobody can prove the model made the call — sunsetsxskies · 2026-07-22
- Richard Hamming’s 1995 lecture asks why mathematics works at all — melnykowycz · 2026-07-22