WaveSeek All channels

Tech · Channel report

What people complain about in AI

WaveSeek read 23,862 public posts about AI over 30 days and found 2,702 genuine complaints in them. Below are the ones that came up most, each linking to the post it was read from.

Measured 17 September 2026 · 12 distinct sources in the last 7 days · this page is rebuilt on every deploy

24,069mentions · 30d
5,534mentions · 7d
2,702complaints · 30d
6competitors measured

Daily mentions, 2026-09-03 → 2026-09-17. Peak 1,113 in a day.

Week over week, mentions moved -7%. A growth figure only appears here when it survives a baseline floor, a source-change check and a credibility ceiling — most channels most weeks show nothing, and that is the honest answer.

The evidence chain

Every complaint below came out of this. Nothing is estimated; each step is a count.

  1. 23,862posts ingested about AI30-day window, after excluding 225 news-wire and video posts
  2. 9,079read by the classifier38% of them
  3. 2,702were genuine complaints30% of what was read

Mentions here exclude news wires and video titles: somebody reporting on AI is not somebody having a problem with it. The Trends screen inside the app counts those too, because it measures attention rather than frustration — which is why the two numbers differ and why this page says which one it is using.

What people actually complain about

The 8 most frequent complaint clusters of 30 found, phrased by the classifier from the posts themselves. The links go to the original post — not to a summary of it.

  1. 1

    Unreliable and inaccurate AI answers

    15 posts across 3 sources · sentiment −0.45

    Why?? — Genuinly useless technology that’s a glorified google search, and it confidedntly lies pretty much everytim…
  2. 2

    Declining accuracy and reliability of AI assistant responses

    10 posts across 2 sources · sentiment −0.52

    The applucation's so useful but it's stuped sometimes 😐👍🏻.
  3. 3

    Restrictive usage limits on AI generation

    9 posts across 2 sources · sentiment −0.36

    Image generation limit — ChatGPT is a really nice app but the time limit is like if you don’t get it like the tournament is so…
  4. 4

    AI assistants lacking persistent memory

    9 posts across 4 sources · sentiment −0.31

    Your AI Has the Memory of a Goldfish. That’s an Architecture Choice. — Most AI assistants are extremely capable and remarkably forgetful. G…
  5. 5

    Unsatisfactory AI assistant quality

    7 posts across 1 source · sentiment −0.79

    Trash — Trash
  6. 6

    Thinking-block errors when reusing conversation context

    6 posts across 1 source · sentiment −0.30

    Claude Fable 5.1 Pensée Conservée : Résoudre l'erreur 'Le bloc est lié à une conversation différente' — Corriger l’erreur de liaison des blo…
  7. 7

    Difficulty navigating long chat conversation histories

    5 posts across 1 source · sentiment −0.16

    Long-Press Conversation Preview with AI Summary — “I’d like to suggest a feature that would make managing a large ChatGPT conversation histo…
  8. 8

    Overly restrictive content moderation limiting usefulness

    4 posts across 1 source · sentiment −0.30

    Heavily moderated — Heavily moderated!!! Cancel your subscription. The extreme moderation makes it practically useless!!! Don’t…

The count is over every post in the cluster. The links are a sample of them, capped at three, and never imply the list is complete.

Who is already there

6 apps measured in this channel from their own recent store reviews. Ratings are computed from the reviews WaveSeek ingested in the window, not read off a store listing.

AppRatingReviews read4★ and up
Google Gemini 4.7★ 200 92%
ChatGPT 4.3★ 200 80%
Grok AI 4.5★ 200 85%
DeepSeek - AI Assistant 2.2★ 136 21%
Claude by Anthropic 3.5★ 128 60%
Perplexity - AI Search & Chat 3.6★ 39 64%

What this page does not tell you

Four competitors in this category print a market size on a page like this one. We do not, because we cannot measure one: mention counts, complaint counts and search volume are not a market size, and a dollar figure built from them would be a guess wearing a decimal point. The same goes for revenue projections, customer demographics and funding comparables.

Everything above is a count of public posts and reviews over a stated window. It is evidence of what people said, not a prediction of what will sell.

See AI change

This page is a snapshot. Inside WaveSeek the same channel is live: the complaint list is re-clustered every six hours, competitor ratings are tracked week over week, and saved ideas report what moved since you saved them.

Open WaveSeek

Free covers the measured numbers you just read. The paid tiers cover the interpretation.