递归智能公司创始人安娜·戈尔迪和阿扎莉亚·米尔霍塞尼认为，更高效的AI芯片设计有助于解决算力危机。创始人周三在TechCrunch Disrupt舞台上详细介绍了他们的初创公司如何开发能够设计芯片、从每次迭代中学习并改进后续设计的AI，目标是实现芯片设计过程的自动化和加速。
Ricursive Intelligence founders Anna Goldie and Azalia Mirhoseini believe more efficient AI chip design could help solve the compute crisis. The founders took the stage at TechCrunch Disrupt on Wednesday to detail how their startup is developing AI that can design chips, learn from each iteration, and improve subsequent designs, all with the goal of automating and accelerating the chip design process.
“我们认为，未来五年对算力的需求将至少增加10000倍，这甚至可能是保守估计，”戈尔迪在舞台上说。“所以我们不想建造10000倍更多的数据中心。我们认为，通过制造更高效的芯片来帮助解决问题——这些芯片能更高效地利用数据中心的带宽，更高效地利用晶圆和晶圆厂，因为即使我们在太空或其他地方建造数据中心，我也不知道，你可能再也看不到太阳了。”
“There is this understanding that in the next five years we are going to have at least 10,000x more demand for compute, and that’s somewhat potentially conservative,” Goldie said on stage. “So we don’t want to build 10,000 times more data centers. We think that we need to help by making chips that are more efficient, that use the bandwidth in those data centers more efficiently, that use the wafers and fabs more efficiently, because even if we make [data centers] in space or something, I don’t know, you won’t see the sun anymore.”
虽然有些人可能对AI自行设计芯片的想法感到担忧，但戈尔迪认为，让AI负责芯片设计将使人类能够专注于开发应用和探索新可能性，从而扩展他们通过计算所能实现的目标。
While some may be scared by the thought of AI designing its own chips, Goldie argues that letting AI handle chip design will allow humans can focus on developing applications and exploring new possibilities that can expand what they can accomplish with computing.
“我认为一切正在发生的转变是：人类贡献他们的伟大想法和创造性思维，而执行由AI来处理，”戈尔迪说。“这正在芯片设计、软件开发和其他领域发生。而递归智能正是在芯片设计领域实现这一点。”创始人表示，AI正在通过发现能够超越传统方案的非常规布局来挑战传统芯片设计方法。例如，他们之前开发的AI系统AlphaChip生成了一种圆形的“甜甜圈形”芯片元件布局，起初让工程师感到惊讶，但最终证明是成功的。
“I think the way everything is shifting is that humans come bring their great ideas and creative thoughts, and execution is handled by AI,” Goldie said. “And this is happening in chip design, software development, and other places. And that’s something that recursive does for chip design.”The founders say AI is challenging traditional chip design methods by discovering unconventional layouts that can outperform traditional ones. For example, AlphaChip, the AI system the founders previously developed, produced a circular, “donut-shaped” arrangement of chip components that initially surprised engineers but ultimately proved to be successful.
他们解释说，他们正在优化整个芯片，而不是遵循传统的做法——将芯片划分为数十个更小的模块。这种方法使人工智能能够创造出复杂的图案，包括类似波浪的设计，而这对搭档表示他们计划在未来分享更多相关内容。
They explained that they are optimizing entire chips rather than following the traditional approach of partitioning them into dozens of smaller blocks. This approach allows AI to create intricate patterns, including wave-like designs, which the pair said they plan to share more about in the future.
创始人还认为，人工智能设计的芯片将使得制造出各种以前可能无法实现的芯片成为可能。
The founders also argued that AI-designed chips will make it possible to create all sorts of chips that might not have been possible before.
“如果设计一款新芯片需要六个月、一年甚至两年，那么你真的必须对所设计的芯片极其挑剔，”戈尔迪说。“但如果你能在一周左右设计出一款芯片，然后用于制造，你就可以为太空、AR、VR以及生物计算等各种不同类型的应用定制专用芯片，而且这样做在财务上也是合理的，我们还能实现更高的能源效率。”
“If it takes six months or a year or two years to design a new chip, then you really have to be extremely selective about the chips that you design,” Goldie said. “But if you can design a chip in like a week, and then say for manufacturing, you can design custom chips for all sorts of different types of applications in like space and AR, VR, and you biological computing, and it makes sense financially to do that, and we can be more energy efficient.”