一位名叫Aldo Cortesi的程序员创造了一种疯狂的算法：一种为不存在的动物绘制轮廓的算法。 其中一些看起来合理，而另一些看起来像来自Alex Garland 2018年科幻电影《歼灭》中的怪异混合体。 (A programmer named Aldo Cortesi has created a crazy algorithm: one that draws silhouettes for nonexistent animals. Some of these look plausible, while others look like uncanny hybrids from Alex Garland’s 2018 sci-fi movie “Annihilation.”)

A few weeks ago we illustrated the possibilities offered by modern neural networks, able to bring back to life ancient times (Denis Shiryaev and his videos).These tools are capable of reconstructing and optimizing images, or, in this case, to build new ones from scratch, thanks to the proper machine learning.

几周前，我们说明了现代神经网络提供的可能性，这些能力可以使古代重生(Denis Shiryaev和他的视频)，这些工具可以重建和优化图像，或者在这种情况下可以构建新的图像从零开始，这要归功于适当的机器学习。

Cortesi wrote that human likenesses-generating algorithms were his main inspiration. While he was mulling the implications of that tech, he found out about PhyloPic, a database that collects silhouettes of animals. He wondered what would happen if he used that cache of data to train a new neural network, designed to create zoological designs.

Cortesi写道，生成人 像是他的主要灵感。 在他研究这项技术的含义时，他发现了PhyloPic ，这是一个收集动物剪影的数据库。 他想知道如果他使用该数据缓存来训练旨在创建动物学设计的新神经网络会发生什么。

Curiosity got the better of him since he was “pretty sure I would get a few good prints for my study wall out of it” — so he adapted existing machine learning code and trained the model using the PhyloPic database. In the end, he generated 50,000 images and sorted through them by hand to find the best examples.

好奇心使他变得更好，因为他“确信我会从学习墙上得到一些好的印刷品” -因此他改编了现有的机器学习代码，并使用PhyloPic数据库训练了模型。 最后， 他生成了50,000张图像，并通过手工对其进行了排序，以找到最佳示例。

The result is an arrangement of organisms that range from the plausible to the nightmarish, a collection of designs enviable even by the best sci-fi designers and 3D artists.

结果是排列了从合理到噩梦的各种生物，即使是最好的科幻设​​计师和3D画家也羡慕不已。

“Straying even further into the unknown, the model produced weird abstract patterns and unidentifiable entities, all with a vaguely biological, ‘life-ish’ feel to them”

“该模型进一步深入未知领域，产生了怪异的抽象图案和无法识别的实体，它们对它们都具有模糊的生物学，'生活化'的感觉”

But he still questions himself about the mechanism behind the result:

但是他仍然对结果背后的机制提出疑问：

“Other aspects of these creatures are fantastically complex and hard to articulate. For instance, what nuances are needed to link the configuration of eyes, mouth, and skin creases into a coherent facial expression?”

这些生物的其他方面异常复杂，难以表达。 例如，需要什么样的细微差别才能将眼睛，嘴巴和皮肤的皱纹的配置与连贯的面部表情联系起来？”

“Of course, I’m anthropomorphizing a statistical machine here, and we may be fooled by our intuition — it could turn out that there are relatively few working variations, and that the solution space is more constrained than we imagine,” he added. “Maybe the most interesting thing is not the images themselves, but rather the uncanny effect they have on us.”

他补充说：“当然，我在这里将统计机器拟人化，我们可能会被直觉所迷惑-事实可能是工作变化相对较少，解决方案空间比我们想象的要受限制，” “也许最有趣的不是图像本身，而是图像对我们的不可思议的影响。”

One thing is for sure: we are only at the beginning of the exploration of the possible results that these AI-led tools can provide us.

可以肯定的是： 我们只是在探索这些由AI主导的工具可以为我们提供的可能结果的开始。

Article by Giulio Vescovi

Giulio Vescovi的文章