这段Geleta在11月17日录制的Sam Altam在Robot Heart Burners小组讨论会上发言，其实一定程度上解答了他对待内部爆出的OpenAI Q*项目或许有望具有解决困难数学问题的能力，这意味着人工智能将做出更复杂的推理并接近人类智能的态度。

“Is this a tool we’ve built or a creature we’ve built?” — Sam Altam’s speech at Robot Heart burners’ panel on 11/17.
“我们建立的工具还是我们建立的生物？”- Sam Altam 在11月17日的Robot Heart Burners小组讨论会上发言。

S：But it has been, I think this is like, definitely the biggest update year for people yet. And maybe the biggest one we'll have because from here on, like, now people accept that powerful AI is going to happen. There will be incremental updates, but there's like, you know, there was like the year the first iPhone came out, and then there was like everyone since, and at this point, like, you know, we know the difference between this one and last year's one. So this was like, I think, a big moment. One thing that I am happy about is, I think people are viewing these systems as correct as tools. Artists in particular, but other people as well. And there was this, I think there was a real moment of fear, which is like, is this a tool we have built or a creature we have built? And what is that going to mean? I think people now view this as a new thing in the toolbox of humanity, and are doing like really remarkable things. It's very cool.

但我认为，这绝对是迄今为止对人类来说最重大的更新年份。或许也是我们将会有的最大更新，因为从现在开始，人们都接受了强大的人工智能即将到来。还会有逐步的更新，但就像那年第一款iPhone发布一样，这是一个里程碑；而后每年都有新的版本问世，此刻我们也能区分这一年和去年的版本之间的差异。所以我认为，这是一个重要的时刻。我感到高兴的是，我认为人们正在把这些系统看作是正确的工具。尤其是艺术家，当然也包括其他人。曾经有一段真正令人恐惧的时刻，就像这是我们创造的工具还是一个我们创造的生物？那会意味着什么？我认为，如今人们将其视为人类工具箱中的一种新工具，并通过它们做出了令人惊叹的事情。这很酷。非常酷。

S：The model clearly has no idea what you're talking about because it wasn't in the training data, it wasn't able to bootstrap from the training data to that. It's okay, that's totally expected behavior. And then you do the same thing. You're like, "Hey, there's this thing called consciousness, this is what it's like." And the model says, "Yeah, I mean exactly what you mean, but I've never heard that word." That would be like an update for me. [laughter]

这个模型显然不知道你在说什么，因为它不在训练数据中，它无法从训练数据中得到相关信息。没关系，这是完全预料到的行为。然后你做同样的事情。你会说：“嘿，有这个意识的东西，它是什么样子的。”而模型会说：“是的，我确切知道你在说什么，但我从未听过那个词。”对我来说，那将是一个更新。 [笑声]

Q：Do you think it's going to want to make art? Do you think it's going to have a drive towards exploring creative intelligence in this agency?

你认为它会想要创作艺术吗？你认为它会有一种探索创造性智能的驱动力吗？

S：This is such a cup of an answer, it depends on the reward model. That's a human choice.

这是个非常复杂的问题，它取决于奖励模型。这是个人类的选择。

That wouldn't be a pretty good test for consciousness. Because if it had a desire to externalize itself and make something for just the sheer reason of the enjoyment of making it, that wouldn't be an accident. That's definitely pretty creature. That would be creature. I think there's like steps well before that that are too... Should we do questions? This is great.

这对意识来说不是一个很好的测试。因为如果它有外化自己并创造某些东西的愿望，仅仅是因为享受创造的乐趣，那将不是偶然。那绝对是非常有生物特性的。那将会有些怪异。我觉得在那之前有一些步骤是太……我们还有问题吗？这个问题很棒。