Xiaomi's Wireless AR Glasses - Tech Insiders News, March 2023
A curated collection of «signal over noise» links about interactive technologies, AI, web3, sense-making, entrepreneurship, and psychology.
- Xiaomi showed off its new AR glasses. Even if not production-ready, it’s a really cool device that pushes the industry forward. A quick primer: most AR Glasses differ from other VR/MR devices like the Meta Quest Pro or the Vive VR Elite: they “just” stream content from your phone.
The Xiaomi ones are wireless, which is an obvious driver of adoption,
And more importantly, these are “real” AR glasses. The other AR glasses on the market overlay visuals in front of you that are not tied to the real world. Here, thanks to the front cameras, virtual objects can be anchored on your environment.
Will AR Glasses be adopted as the “second screen” for smartphones in the future? It might be, as it’s lighter and cheaper than full VR/MR devices. - Last week, in just a few days, Open AI, the world’s leading AI company, published a statement about AI Safety and released an API for ChatGPT at a 10th of the price of the previous models.
First, why would AI pose existential risks to humanity?
Let’s use an example: chess. Let’s assume that there is an AI who’s better than all the chess masters. They cannot predict that AI's next move: if they could, that AI will not be better than them, right?
Now, assume that there is an AI with the same quality (better than all humans) but in every aspect of humanity. Games, economics, history, technology, biology, war… This is what we call a super-intelligent AI.
Then, by definition, we cannot predict what this AI will do. If we could, it would not be better than all humanity.
And because we cannot predict what it will do, then… We have to imagine the worst. What would be its goals? How would it achieve them? What side-effect will appear during the process of achieving these goals?
This is what Open AI is addressing in its blog post. Basically, they believe that a gradual transition with incrementally more powerful AI that we can correct along the way is the best way to go.
But then, a few days later, they release the ChatGPT API for a cheap price. This will allow deeper integration of ChatGPT in many software, as it was only a web UI up to now. Am I wrong in seeing incoherence here? I don’t think we are yet adapted to ChatGPT and Bing Chat… Brace yourselves for a flourishing of new applications in the coming weeks.
Consider what you are doing at work or for a hobby, and imagine how conversational AI could be applied to it. I strongly believe that it will come for every one of you.
Yet, I would like to nuance a bit: even if not optimal, the fact that Open AI is planning for AI safety is a good thing!
Dive Deeper: A long analysis of OpenAI’s post and an interview with Eliezer Yudkowsky, the thought leader on AI risks. - Many libraries used in the code of websites are free and quite often created and maintained by only one person. And sometimes, this library is used in so many projects that it becomes essential. What if its creator cannot keep on maintaining it? This is what is currently happening with CoreJS.
This had been a wake-up call for many of us. At Gimbal Cube, we’ve made a recurring donation. If this resonates with you, go to this link to contribute, and let me know if you did!
Today’s image, thanks to Midjourney: “A benevolent AI broom“.