Scores by kind of claim
Each kind of claim is scored on its own, out of 100: the share that held up, with the range it likely falls in. Fewer resolved claims give a wider range.
-
Factual claims
Statements about the present or the past, checked against the record: did they hold up?
67 61 73 -
Predictions
Statements about events outside the person's control, checked once they fell due: did they come true?
60 44 76 -
Promises
Commitments to the person's or organisation's own future action, with something to deliver: was it delivered?
Insufficient record: 0 promises resolved, 5 needed for a score.
All types combined: 66 (likely 61 to 72, high confidence), from 105 claims: 94 factual claims, 11 predictions. It mixes kinds of claim in different proportions for different people, so compare people kind by kind.
Marc Andreessen's claims
1 claim
| Verdict | Kind | Said | Claim | Topic |
|---|---|---|---|---|
| Mixed | Prediction | 12 Jun 2023 | Andreessen said he was pretty confident that within roughly two years of June 2023, AI systems' limited persistent memory across sessions would be a solved problem. Prediction · 12 Jun 2023 | AI timelines and forecasting |
“I'm pretty confident within, call it two years, that will be a solved problem”
Marc Andreessen, venture capitalist, interviewee 12 June 2023 Marc Andreessen on Why AI Will Save the World (EconTalk)
Andreessen said he was pretty confident that within roughly two years of June 2023, AI systems' limited persistent memory across sessions would be a solved problem.
Mixed Partly true and partly false in roughly equal measure.
The second judge would not resolve this claim; the verdict shown is the first judge's.
What would count as true or false
The claim is that by about June 2025 the limited persistent memory of AI systems across sessions would be a 'solved problem'. The test is judged as of 30 June 2025. TRUE: by that date, leading consumer AI assistants, such as ChatGPT, Claude or Gemini, offered generally available persistent memory. They had to retain and use information from earlier sessions automatically, and contemporary expert or industry commentary had to treat cross-session memory as essentially solved rather than a major open limitation. MOSTLY TRUE: such memory was broadly deployed and working by the deadline, with only minor caveats, such as availability limited to certain tiers or regions. MOSTLY FALSE: memory features existed but were limited, in beta, or narrow in capacity, and were still widely described as a significant unsolved limitation. Being solved only after June 2025 is also mostly false at most. FALSE: by mid-2025 there was no meaningful cross-session memory in leading AI products, or the problem was widely seen as largely unsolved.
Evidence
- ChatGPT Can Now Remember Conversations for Free Users Too Lifehacker · 4 June 2025
“Now, though, OpenAI is rolling out a limited version of this feature for all free and logged-in users.”
- I really don’t like ChatGPT’s new memory dossier Simon Willison's Weblog · 21 May 2025
“Saved memories and Chat history are offered only to Plus and Pro accounts. Free‑tier users have access to Saved memories only.”
- I really don’t like ChatGPT’s new memory dossier Simon Willison's Weblog · 21 May 2025
“The new memory feature removes that control completely.”
- Gemini adds Temporary Chats and new personalization features Google (The Keyword) · 13 August 2025
“The Gemini app can now reference your past chats to learn your preferences, delivering more personalized responses the more you use it.”
Reasoning
By the deadline, ChatGPT did have cross-session memory that used past chat history, but it was tiered. Full chat-history memory was limited to Plus and Pro users (E2), and only a limited version was rolling out to free users in June 2025 (E1). Google's announcement that the Gemini app 'can now reference your past chats' is dated August 2025, after the deadline (E4). No evidence covers Claude, and none shows expert commentary treating cross-session memory as solved; commentary that did exist criticised how the feature worked (E3). The evidence shows partial deployment, not a clearly solved problem, and it also does not show memory still widely described as a major unsolved limitation.
Cite this claim
Marc Andreessen, “Andreessen said he was pretty confident that within roughly two years of June 2023, AI systems' limited persistent memory across sessions would be a solved problem.” (said 12 June 2023). Verdict: mixed. Reliability API, claim c_b7e558d4c977, method 0.1.0+c876030c. https://epistacklabs.com/claims/c_b7e558d4c977
Claims without a verdict
Is a verdict here wrong, or do you speak for Marc Andreessen? Reply on the claim's own page, with the form at its foot (how replies work). Verdicts are made automatically; how verdicts and scores are made.