Highest paid and most powerful people in human societies are those who manipulate other humans, and CEOs are in this category. They're not being paid so much to make decisions, they're paid for their connections. Taking that into account, AI would be a powerful tool for CEOs as they can level the playing field for the ones who keep "fail upward".
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
This universe appears to have the property that increasingly complex information processing capabilities will appear in exponentially decreasing time, with new more complex substrates for information processing created by the previously most complex substrate.
Minimax is one of the OG big 4 up and coming AI startups tiger of China along with Zhipu (GLM) Moonshot (Kimi) and Baichuan.
However it was originally more famous for its Hailuo Ai video generation that was free for 5 second clips. Sites like 4chan used it a lot to originally make funny stuff and later racist stuff. One of the more infamous ways it's product spread. That and it's Talkie AI character chat bot.
So the fact they can create an LLM that kicks almost everyone's ass despite it's speciality being something else shows the potential of minimax.
The head of MiniMax, Yan Junjie, said recently that he expects the costs associated with AI inferences to drop by 10x within two years. The MiniMax-M1 already performs significantly more efficiently than DeepSeek, so I believe him.
Not to go too far off topic, but one thing I don't necessarily believe is when Yan says he believes there is enough room for growth that a lot of companies can coexist without a Warring States-like battle for supremacy. Nothing about any industry's corporate history supports that, and I don't think any CEO has ever not wanted to have total domination over their competitors. I am not faulting him for that, but I just don't believe it.
So where Yan says open-source models and multi-agent systems make it unlikely for any single company to dominate the market, I counter that open-sourcing engines and tools is a temporary tactic in service of the long-term goal of dominating the market. Once users become dependant on certain tools or have fuly bought into their ecosystems, newer models will likely either stop being open sourced, or some other even more devious cost-extraction measure will be leveraged that we have not yet thought of. Again, not a fault of any one company, but companies are not public services.
If you look back at almost every major company with a strong brand identity, you observe a life-cycle in their public perceptions. They start out bright and shiny and can do no wrong. Then somewhere along the road they do something really shitty or something really shitty that they did in the past gets exposed, and everyone turns against them but its already too late because they have a monopoly. What is the takeaway lesson? Never feel too warmly toward any company.
You bring up good points, actually think they are great points, but I suspect there will be a flood of different models, like what that CEO guy said.
Why?
Let's ask ourselves this question. How many computer programming languages are there? I don't know, last time I read about it, years ago, there were like 3000 languages. How many are still in current use? I forget, probably 100 of them?
Even Cobol is still used, and that is like twice the age of the average poster in this forum!
Remember Ma Yun and his micro-loans? That was from the algorithm and that was AI. But his AI to make small loans, has to be different for a bio-tech company using AI in a lab developing new drugs, which has to be different from the AI in a car factory when they have to weld two pieces of metal together, which has to be different from logistics AI of transporting goods, from agricultural production as they used AI to recognize cows I remember watching in a video years ago.
Why is there so many computer languages? Like 3000 of them. A countless numbers of compilers! The geeks gotta to geek probably is the best explanation.
With AI, which is going to cover everything, we know there will be a niche everywhere and not all models will be doing the same things. We already have video making models and coding models, etc.
Unless someone builds the God Model of LLM, then I suspect the open source will flourish with many developers across many industries.
Again, we come back to that same question, at least in my mind, of where the world stands with AI today, (which basically two countries walking that path through the forest).
It is all about AGI the Artificial General Intelligence, aka the God Mode or Skynet, and the diffusion.
Essentially what that CEO is really saying, is diffusion will create a lot of opportunities for a lot of companies.
Did he say anything about the God Mode? Did not read that piece. Theoretically that is possible, one God Mode AGI can dominate the market.
Maybe AI will be like the computer languages. Too many of them.
This is kind of cool. Hunyuan's 3D model generation AI model has been combined with 3D printing to create all sort of little creatures.
Another huge data center getting built in China. This time 65 EFLOPS with 11.8B RMB investment in Huizhou. China's regional data center buildout is so much larger than people can appreciate.
I just returned from a business trip, and my development is mainly focused on embedded environments, as well as tinkering with NPU AI applications. The solutions I have been exposed to include
rk3399/rk3568/rk3576/rk3588/ax630/ax650（1-18TOPS int8）
What I am doing in the future will basically involve the targeted application of all domestic AI SOC, including those discussed by the general public and those not discussed
I haven't had much exposure to AI SOC now, purely because I don't have the money to buy too many development boards and carry out development (I am currently working on mass production customization of AI hardware, with an estimated investment of 120000+USD), because I will add NPU/AI functionality to a specific customized Linux solution. It is 10 times more troublesome than on official custom Linux distributions (usually Ubuntu/debian/armbian).
I'm no amateur in AI hardware/software, though my projects primarily use the YOLO series—not LLM-type models. Truth is, current-generation NPUs like those in the RK33/RK35 series SOCs aren’t well-suited for LLMs.
It’s the next-gen RK36 series that’s designed for it.
I respect the information shared by the person in the video, though I wouldn’t consider them an expert either. That said, their insights are particularly intriguing.
I’ve engaged with acknowledged heavyweights in China’s internet circle—many now diving into AI. Technically, I’m confident we have a far clearer grasp of the underlying tech than these industry leaders.
But commercially speaking, early-stage innovation ventures’ survival—fundamentally reliant on a CEO’s fundraising prowess—means individuals like those in the video prove indispensable in the commercialization process.
Last edited: