Click the card below to follow the " CVer " official account
AI/CV heavy dry goods, delivered in the first time
Click to enter —>【Computer Vision】WeChat Technology Exchange Group
Reprinted from: Xin Zhiyuan | Editor: La Yan
[Guide] Turing Award winner Yann LeCun talked about AI: the future is open source.
The first half of this year can be described as the most turbulent half year for the AI industry.
Behind the rapid development of various GPTs and even the prototypes of AGI are people from two camps with different views.
One faction believes that the generative AI headed by ChatGPT is very powerful and can drive a wave of revolutionary trends, and there is no problem in continuing to advance.
The other school of thought thinks that we are growing a little too fast. Not to mention prohibition, but also to stop. And morally, there are too few matching constraints.
As a big player in AI, Yann LeCun has a different view on this.
Outright ban? not feasible
LeCun said he wasn't surprised by ChatGPT's performance, nor was he in favor of a moratorium on AI research.
"This could have the opposite effect than expected."
He said that artificial intelligence, as an amplifier of human intelligence, may be the origin of the new renaissance.
The large language model of ChatGPT is "autoregressive". The AI is trained to draw words from a corpus of up to 1,400 billion words to predict the last word in a given sequence of sentences, which must occur next.
The related research carried out by Claude Shannon in the 1950s was based on this principle.
The principle has not changed, it has become the size of the corpus, and the computing power of the model itself.
LeCun said, "Currently, we cannot rely on such models to generate long and coherent texts. These systems are not controllable. For example, we cannot directly ask ChatGPT to generate a text whose target audience is 13-year-old children.
Second, the text generated by ChatGPT is not 100% reliable as an information source. GPT functions more like an auxiliary tool. Just like the existing driver assistance system, you have to control the steering wheel when you turn on the autopilot function.
Moreover, the lifespan of the autoregressive language models we are familiar with today is very short, and five years is considered a cycle. After five years, no one will use the past models anymore.
The focus of our research should be on finding a way to make these models controllable. In other words, the AI we want to study is an AI that can reason and plan according to a given goal, and must be able to ensure that its safety and reliability standards are consistent. This AI can feel emotions. "
You know, a large part of human emotions is related to the achievement of goals, that is, to some form of anticipation.
With such a controllable model, we can generate long and coherent text.
LeCun's idea is that in the future, he will design enhanced models that can mix data from different tools, such as calculators or search engines.
Models like ChatGPT are only trained on text, so ChatGPT's knowledge of the real world is not complete. And if you want to develop further on this basis, you need to learn something related to the sensory perception and world structure of the whole world.
And these more complex contents can not be realized simply by reading the text, which is one of the biggest challenges in the next few years.
open source is the end
The desire for power is unique to human beings. And AI does not have this desire just because it becomes more and more powerful.
Only the human species knows how to formulate laws to ensure that individual actions will not damage the common interest too much.
OpenAI started as an open research project, which is now closed. While OpenAI has said nothing about its work, this reversal of the situation is uncommon in the research world.
The problem is that training a language model is expensive, costing tens of millions of euros, so startups cannot afford it.
This is also the main reason why Microsoft merged with OpenAI, requiring the group's collective computing power to improve its future models. That's why DeepMind and Google Brain eventually merged.
LeCun said that in the end, in terms of the market, developers will move towards a common ecology of an open platform. It would be bad if only a few companies controlled the technology.
Historically, both Facebook and the renamed Meta have been actively promoting open basic research, such as the open source project LlaMa.
In the early 1990s, Sun Microsystems and Microsoft fought each other for the right to operate servers. Remember, all Internet technologies that have always stood firm are open source.
LeCun finally said that at present, the key to preventing such open source AI platforms is legal issues. Such an open source platform is essential if the EU wants to promote the structuring of the AI industry in the future.
References:
https://twitter.com/USBEKetRICA/status/1648597311843450881
Click to enter —>【Computer Vision】WeChat Technology Exchange Group
The latest CVPR 2023 papers and code download
Background reply: CVPR2023, you can download the collection of CVPR 2023 papers and code open source papers
Background reply: Transformer review, you can download the latest 3 Transformer review PDFs
目标检测和Transformer交流群成立
扫描下方二维码，或者添加微信：CVer333，即可添加CVer小助手微信，便可申请加入CVer-目标检测或者Transformer 微信交流群。另外其他垂直方向已涵盖：目标检测、图像分割、目标跟踪、人脸检测&识别、OCR、姿态估计、超分辨率、SLAM、医疗影像、Re-ID、GAN、NAS、深度估计、自动驾驶、强化学习、车道线检测、模型剪枝&压缩、去噪、去雾、去雨、风格迁移、遥感图像、行为识别、视频理解、图像融合、图像检索、论文投稿&交流、PyTorch、TensorFlow和Transformer等。
一定要备注：研究方向+地点+学校/公司+昵称（如目标检测或者Transformer+上海+上交+卡卡），根据格式备注，可更快被通过且邀请进群
▲扫码或加微信号: CVer333，进交流群
CVer计算机视觉（知识星球）来了！想要了解最新最快最好的CV/DL/AI论文速递、优质实战项目、AI行业前沿、从入门到精通学习教程等资料，欢迎扫描下方二维码，加入CVer计算机视觉，已汇集数千人！
▲扫码进星球
▲点击上方卡片，关注CVer公众号It's not easy to organize, please like and watch