2023年经济学人 新的联想学习模型与其对人工智能的影响(2)（在线收听）

    Looking at things this way deals with two things that have always bugged the old model.

    以这种模型看待问题解决了两个旧模型无法解释的问题。

    One is sensitivity to timescale.

    一是时间上的敏感性。

    The other is computational tractability.

    二是计算上的可操纵性。

    The timescale problem is that cause and effect may be separated by milliseconds (switching on a light bulb and experiencing illumination), minutes (having a drink and feeling tipsy) or even hours (eating something bad and getting food poisoning).

    时间方面的问题在于，因果可能以毫秒（打开灯泡并感受到照明）、分钟（喝饮料并感到醉意）甚至小时（吃坏肚子并食物中毒）为单位出现。

    Looking backward, Dr.Namboodiri explains, permits investigation of an arbitrarily long list of possible causes.

    南博迪里博士解释说，回顾过去，可以分析很多可能造成某个结果的原因。

    Looking forward, without always knowing in advance how far to look, is much trickier.

    但展望未来，不总是能够知道要等待什么时候发生的什么结果，分析起来要困难得多。

    This leads to the second problem.

    这就导致了第二个问题。

    Sensory experience is rich, and everything therein could potentially predict an outcome.

    感官体验是十分丰富的，其中一切都有可能是结果的预测。

    Making predictions based on every single possible cue would be somewhere between difficult and impossible.

    根据可能的单一线索进行预测十分困难，甚至不可能实现。

    It is far simpler, when a meaningful event happens, to look backwards through other potentially meaningful events for a cause.

    而如果发生了一件有意义的事，据此回顾其它潜在的、有意义的事来寻找原因要简单得多。

    In practice, however, it is hard to distinguish experimentally between the two models.

    然而现实是，很难在实验中区分这两种模型。

    And that is especially true if you do not even bother to look - which, until now, people have not.

    如果你甚至懒得费心去看，那就更是如此——到目前为止，还没有人做过这样的事。

    Dr. Jeong and Dr. Namboodiri have done so.

    而郑博士和南博迪里博士做到了这一点。

    They devised and conducted 11 experiments involving mice, buzzers and drops of sugar solution that were designed specifically for the purpose.

    他们设计并进行了11项实验，实验用到了小鼠和专门为实验设计的蜂鸣器和糖液。

    During these they measured, in real time, the amount of dopamine being released by the nucleus accumbens, a region of the brain in which dopamine is implicated in learning and addiction.

    伏隔核是大脑中与学习和成瘾有关的区域，在实验过程中，研究人员对伏隔核释放的多巴胺含量进行了实时测量。

    All of the experiments came down in favour of the new model.

    所有的实验结果都支持新的模型假设。

    The 180° turnabout in thinking - from prospective to retrospective - that is implied by these experiments is causing quite a stir in the world of neuroscience.

    这些实验所暗示的思维180度大转变——从前瞻性思维到回顾性思维——在神经科学界引起了不小的轰动。

    It is "thought-provoking and represents a stimulating new direction," says Ilana Witten, a neuroscientist at Princeton University uninvolved with the paper.

    没有参与这篇论文的普林斯顿大学的神经学家伊兰那·维登说，这篇论文“发人深省，代表了一个鼓舞人心的新方向”。

    More experiments will be needed to confirm the new findings.

    还需要更多的实验来证实新的发现。

    But if confirmation comes, it will have ramifications beyond neuroscience.

    但如果该发现的真实性得到证实，它产生的影响将不仅限于神经科学。

    It will suggest that the way AI works does not, as currently argued, have even a tenuous link with how brains operate, but was actually a lucky guess.

    它表明，人工智能的工作方式并不像目前人们所说的那样，与大脑的运作方式有着微妙的联系，而实际上是幸运的猜测。

    But it might also suggest better ways of doing AI.

    但它也可能为人工智能学习提供更好的方法。

    Dr. Namboodiri thinks so, and is exploring the possibilities.

    南博迪里博士就是这样认为的，并正在探索新的可能性。

    Evolution has had hundreds of millions of years to optimize the process of learning.

    数亿年的进化在不断优化学习的过程。

    So learning from nature is rarely a bad idea.

    因此，向大自然学习不失为一个好主意。