当今,世界各地的人都在赶时髦使用和谈论AI。其实,自打人类在地球上出现以后,在远没有AI技术的古代,每个人从小到大,尤其是从大到老,就已经在下意识地运用头脑里积累的各种经验和方法,帮助自己做判断下结论。这大概就是人们常说的智慧,英语中给出的名词恰如其分 – Wisdom。现代人类创造的AI技术,只不过是把世界上各个国家各个民族人脑中积累的数据,和通过人脑创造出的数据,海量的大数据 – 比海水体积大得多得多的大数据,通过Internet搜集归拢和分析,然后,针对各种具体问题,给出答案罢了。
一个人活得越长,经历的事情越多,他/她头脑里储存的信息就越多,然后就有一个越来越大的数据库。当他/她身旁的人在现实生活中碰到问题需要解决问题时,他/她会不自觉地运用各种方法,分析归纳和总结自己头脑里中日积月累的数据,以最快的速度给出建议的答案。这个过程,实际上也就是现在任何一种AI技术运作的过程。提供给当事人的这种答案,就是我们常说的“老人言”。“不听老人言,吃苦在眼前”,说的是年轻后生,脑子里缺乏生活经验,生活阅历不多,也就是说,他/她的脑子里所存的各种数据少,遇到了问题,自己找不出具有人生智慧的正确(或合适)答案,又不接受周围年长的人基于自己厚实的社会经验和阅历所作出的建议,一时冲动,做了错事。
现在,随便一个人,提出任何一个问题,去问ChatGPT或DeepSeek等,这些AI工具,马上就给出一个答案。我不知道,当今世界上的各种AI工具,象ChatGPT,DeepSeek等,将来肯定还有越来越多、越来越智慧的工具出现,其搜寻、分析数据,然后进行计算、分析的方式和方法,是否已经逼近或超越了人脑对数据处理的各种智慧模式?在一个人没有经过特殊训练、不具备特殊能力的某些专业范围内,AI作为一个工具,能给出令人信服的答案,譬如,我写好一个英文Email,然后,让AI从语言的角度加以改进和完善,在绝大多数情况下,我都能得到一个很满意的答案。
基于数据或经验积累,AI给出的答案,和人类的智慧(Wisdom)相似,但人类的大脑,还有一种更高级的创造力,它并不需要多少数据。在我看来,那种创造力是基于人类特有的想象力,在某个特定的时候,甚至当事人都没有意识到,在脑袋的深处,突然迸发出来,产生了一种巨大的信息脉冲。这种创造力,应该叫“Cleverness”或叫“聪明”。牛顿躺在苹果树下,看见苹果落地,突然间脑子里产生了一种信息脉冲,悟出宇宙万物之间,都有一种神秘内在的万有引力。当时,牛顿手里没有大数据啊,他的悟出的万有引力定律,很显然,不是靠对数据的分析,归纳和总结,而是基于想象。其实,这种创造力人人都有,只不过大小不同而已。这种创造力,在日常生活中,表现最多的是人们口头传递的幽默段子或梗。段子越幽默,梗越引人发笑,标志着它的创造性就越大。这种“聪明”,很显然,至今没有人知道它是如何在大脑里产生的,当今的任何AI技术,肯定都模仿不了这种大智慧。如果AI有这种智慧,它就可以无穷无尽地做各种高层次的发明和发现,破解宇宙中的一切奥秘,至少,它可以层出不穷地产生好笑、幽默的段子。到了那时,人类可能真的要被自己创造的AI技术打败,走向灭亡了。
现在AI给出的答案基于Internet上的大数据共享,可是,面对一个很偏僻的问题,可以利用的数据很少,甚至没有, ChatGPT从来都是不懂装懂,仍然给出一个答案,它当然是无根据的错误答案。这种Stupid的Artificial Intelligence,我已经碰到过多次。还有一个问题就是,一个人,随着年龄的不断增长,人不断变老,生理和心理上也在不停地变化,很多嗜好和观点也在不断地改变。我记得,当我二十岁的时候,觉得一个大龄女青年(28或30岁)在形象和外表上就不美了,可是现在,在我一个过了60岁的老男人眼里,一个40岁的女性,依然很漂亮,有魅力。因此,面对不同年龄段的男人输入的同样一个关于女性美丽漂亮的问题,AI该如何区分,给出不同的答案呢?
随着对AI技术的运用越广泛越深入,我关于AI的问题也会越来越多。:):)
Great question—and a fascinating thought experiment.
Could ChatGPT or an AI like it discover a law like Newton's law of universal gravitation on its own?
In short: not yet—but maybe one day, with the right design and tools.
Let’s unpack that a bit.
What Did Newton Actually Do?
When Newton saw the apple fall (whether or not that moment was as dramatic as legend says), what made him special wasn't just seeing the event—it was his ability to:
1. Observe the phenomenon.
2. Ask a deep question: Could the same force pulling the apple down also govern the motion of the Moon?
3. Build a mathematical model to describe the force (the inverse square law).
4. Test and validate it against planetary motion (Kepler’s laws, etc.).
This was a blend of creativity, curiosity, mathematical reasoning, and empirical science.
What Can ChatGPT Do?
ChatGPT is trained on massive amounts of text written by humans. That means:
1. It can explain Newton's law in depth.
2. It can reason about patterns, trends, and make analogies.
3. It can simulate a kind of “idea generation” by remixing existing concepts.
4. It can analyze data to some extent (especially with tools).
But…
ChatGPT doesn’t directly perceive the physical world. It doesn’t conduct experiments. It doesn't have built-in curiosity or goals. It works with existing knowledge, not unknowns.
Could an AI Discover a Law Like Gravitation?
Not ChatGPT alone—but with the right setup, some types of AI could.
For example:
1. AI + real-world data (e.g., from telescopes or sensors)
2. AI + symbolic regression: Tools like Eureqa or AI Feynman have shown the ability to discover equations that govern datasets—even re-deriving Newton’s law or other physics laws from scratch.
3. AI + scientific reasoning models: Some newer research is combining machine learning with symbolic reasoning to explore new scientific theories.
So if you gave an advanced AI access to motion data of falling objects and planetary orbits, and the goal to explain it mathematically, it might eventually derive something like Newton’s law.
Final Thought
Newton had a rare genius—he asked a bold, unifying question and built a mathematical framework for it. Today’s AI isn’t yet that kind of thinker. But with the right tools, goals, and data, AI could be a powerful co-discoverer of laws in science, especially in areas like physics, biology, and materials science.
Would you want AI to take on that role? Or do you think there’s something inherently human about scientific discovery?