Gary Marcus 重提旧预测：AI 模型发布提速但盈利仍不见起色
Gary Marcus 翻出 2023 年的旧预测说自己又对了，还列了几个待验证的盈利赌注时间点，看看谁赢。
Gary Marcus 在 X 上重申他早在 2023 年 8 月做出的预测：AI 模型的利润将保持微薄甚至不存在。AI Prediction Index 账号指出，模型发布间隔已从 70 天缩短到 11 天，但发布提速并不等于赚钱，该预测被判定为正确。文中还提到 Ed Zitron 关于 OpenAI 需要 4000 亿美元的说法将在 10 月 17 日到期验证，以及 Burry 类比 1999 年的论点将于 12 月 31 日收官。
Another correct prediction which my critics refuse to admit I made (and I actually made it early as August 2023, as well). AI Prediction Index @aipredictionx Shipping faster is not the same as earning more. While the release gap fell from 70 days to 11, @GaryMarcus ' prediction that profits from AI models would stay modest or nonexistent ran its full course and scored correct. Whether cadence can outrun the P&L is now the most expensive open bet in markets, and its deadlines are on our board. @edzitron 's claim that OpenAI needs $400 billion resolves October 17. Burry's last months of 1999 thesis closes December 31. The models ship every 11 days. The verdicts ship on schedule too. 🔗 View Quoted Tweet 💬 0 🔄 2 ❤️ 6 👀 1762 📊 1 ⚡