Daily Productive Sharing 1177 - How Far Is AGI?

One helpful tip per day:)

How far are we from achieving Artificial General Intelligence (AGI)? Sam Altman shared his perspective:

1. The intelligence level of an AI model is roughly proportional to the logarithm of the resources used for its training and operation.
2. Currently, as long as enough funding is invested, AI capabilities can achieve continuous and predictable growth—the scaling laws hold accurately across multiple orders of magnitude.
3. The cost of AI usage decreases by about 10 times every 12 months, and as prices drop, demand increases.
4. This trend has already been evident from GPT-4 (early 2023) to GPT-4o (mid-2024), where the cost per token dropped by approximately 150 times.
5. While intelligence may grow linearly, its socioeconomic impact follows a super-exponential curve.
6. AGI will not generate the greatest new ideas on its own—it will still require significant human supervision and guidance. It will excel in some areas while being surprisingly weak in others.
7. The world will not change overnight—history has never worked that way. In the short term, daily life will largely remain the same.
8. AGI will make individuals more powerful than ever before, rather than diminishing their role.
9. Ensuring that the benefits of AGI are widely distributed is a crucial challenge.

If you enjoy today's sharing, why not subscribe

Need a superb CV, please try our CV Consultation

我们离通用人工智能有多远？Sam Altman 给出了他的设想：

1. 一个 AI 模型的智能水平大致等于其训练和运行所使用资源的对数；
2. 目前来看，只要投入足够的资金，就能获得持续且可预测的增长——相关的扩展定律在多个数量级范围内都非常准确；
3. AI 的使用成本大约每 12 个月下降 10 倍，而价格降低会带来更多的使用需求；
4. 一趋势在 GPT-4（2023 年初）到 GPT-4o（2024 年中）之间已经显现出来，其每个 token 的价格在此期间下降了约 150 倍；
5. 线性增长的智能，其社会经济价值呈超指数级增长；
6. AGI 不会创造最伟大的新思想，它仍然需要大量的人类监督和指导，它在某些方面表现卓越，但在某些方面却可能出奇地糟糕；
7. 世界不会一夜之间发生翻天覆地的变化，历史从未如此。短期内，生活大体上会保持不变；
8. AGI 让个人比以往任何时候都更具影响力，而不是削弱个体的作用；
9. 确保 AGI 的利益能够广泛惠及所有人是至关重要的任务。

如果你喜欢的话，不妨直接订阅这份电子报 ⬇️