AI Era Network Comprehensive Report丨【June 19, 2023 Morning Post】
Bill Gates speaks on AI: the second revolutionary turn of life
Bill Gates admits that he is witnessing "the second revolutionary moment of his life," writing on his blog: "The Age of AI has begun."
He believes that revolutionary innovations in artificial intelligence will revolutionize the way people work, study, travel, treat and communicate, just like PCs, the internet and mobile phones, and even reduce some of the world's worst inequities.
Bill Gates said he sees the "first revolutionary moment" in 1980, when he was exposed to the graphical user interfaces of computers, including Windows and Mac, which we all use today. He said he met Charles Simeone, a talented programmer, and the two sat down together to brainstorm all the scenarios that could be used graphically in the future. Simeone eventually joined Microsoft, and Windows became Microsoft's pillar industry, and the thinking at that time determined the company's development direction for the next 15 years.
The second revolutionary moment that Gates is talking about is last year, 2022, and in fact, Gates has been meeting with the OpenAI team since 2016. He said that during a meeting in mid-2022, he was very excited about OpenAI's work achievements. He presented a challenge: To train an AI to pass a pre-college biology exam, the AI must be able to answer questions that it wasn't trained in.
AI is frantically polluting Chinese internet
AIGC is widely used, but a terrible fact is gradually revealing, and AI is polluting the entire Chinese Internet.
Under the AI-compliant organization, meaningless content has skyrocketed, and fake news and headline parties have gained a lot of exposure. AI not only does not liberate productivity, but bad money drives out good money, makes good creators lose enthusiasm, and also creates a more serious information cocoon.
At the same time, Zhihu has become the hardest hit area for ChatGPT to generate brainless answers. On Zhihu, some people asked if there were any interesting or suspenseful true stories of ancient China. Someone immediately answered, the Southern Ming thief Huang Zongxi.
Someone asked AI if there was a cable car in Elephant Trunk Mountain, and after the AI answered in the affirmative, it also thoughtfully attached detailed information such as fares and business hours. However, this netizen still holds a puzzled mentality and clicks the reference link behind. Netizens felt how strange this account was, so he clicked on the user homepage called "Variety Life", and suddenly found that this account is an AI!
But in fact, this phenomenon of "AI pollution sources" does not only exist in China, but also abroad. AI models obtain large amounts of Internet data, but cannot distinguish the authenticity and credibility of the information well.
We ended up having to deal with a whole bunch of low-quality content that was generated quickly, making people dizzy and dizzy. It's hard to imagine big models like ChatGPT being trained on this kind of data.
Microsoft partnered with Mercedes-Benz to integrate ChatGPT into the vehicle system
In the ever-evolving field of automotive technology, the Mercedes-Benz automaker became the first company to integrate ChatGPT into its infotainment system. ChatGPT is a conversational AI agent developed by OpenAI, which has extensive language understanding and generation capabilities to have natural conversations with users on a variety of topics.
Mercedes-Benz announced this week that their MBUX infotainment system will begin beta testing, which integrates ChatGPT. The test program will be available to select U.S. owners on Friday, June 16 and covers more than 900,000 MBUX-equipped cars, covering 27 model ranges from the A238 to the Z296.
Wang Xiaochuan launched a free commercial large model
After two months, the "Baichuan Intelligence" formed by Wang Xiaochuan officially launched the first 7 billion parameter Chinese and English pre-training large model - baichuan-7B on June 15. The baichuan-7B not only surpasses other large models such as ChatGLM-6B in the C-Eval, AGIEval and Gaokao Chinese authoritative evaluation lists, but also significantly ahead of LLaMA-7B in the MMLU English authoritative evaluation list. The baichuan-7B model has been released on Hugging Face, Github, and Model Scope.