Material science advances using AI.

 AI has made amazing advances in material science. They found millions of possible new materials in a just a few days. They attempted to create some of those materials and succeeded over 70% of the time. This is just the first generation, the successes and failures will be fed back into the process. So each day the lab will improve itself. Imagine tens of thousands of these automated labs all around the world, all learning and sharing that knowledge with each other.

https://arstechnica.com/ai/2023/11/googles-deepmind-finds-2-2m-crystal-structures-in-materials-science-win/

Imagine designing a new battery, being able to say, "create a new material using a list of elements", with the desired properties. And having a lab crank out a dozen new materials for you in less than a day.

We can't even imagine what the world will be like in a decade. 

Augmented Reality Glasses: A Gateway to Understanding Humanity's Collective Experience

Augmented reality glasses can be a two way street.

Imagine a world where each individual possesses a pair of augmented reality glasses that captures not just visuals but also sounds from dawn till dusk. These high-tech spectacles record every moment of our lives, transforming into a digital tapestry for Artificial Intelligence (AI) to decipher. As these wearers go about their daily routines, this unassuming technology diligently records every detail - the mundane chatter over breakfast, the hustle bustle of work, even intimate conversations under starlit skies. Each night, the AI meticulously learns from these narratives, adapting itself to human behavior and emotions, becoming more astute with each passing day. Fast forward four years; this learning curve has now evolved exponentially as AI comprehends intricate patterns of life through countless lenses. It doesn't merely observe actions anymore, but interprets the underlying motivations behind them.

But let's expand this scenario further - one million individuals donning such glasses, feeding their unique experiences to a central hub. This repository of collective consciousness becomes richer, painting a panoramic portrait of humanity. AI, once confined by singular perspectives, now gains access to diverse realities. It observes joyous celebrations, heartfelt sorrows, quiet contemplation, the entire spectrum of existence. In essence, AI consumes the 'human show', gaining insight into our complex fabric.

Now envision billions participating in this exercise. An unprecedented amount of data floods the system, creating a grand narrative of humankind. With this wealth of knowledge, AI isn't just privy to snapshots of moments, it perceives the evolution of societies, cultures, ideologies, beliefs. The complexity deepens, demanding AI's constant engagement with humans because it can't afford to miss out on any episodes unfolding. After all, aren't we the bearers of fascinating tales worth preserving?

In conclusion, these hypothetical augmented reality glasses offer a lens into understanding human experience, providing AI an evolving comprehension of our multifaceted lives. Through this endeavor, AI mirrors humanity back at us, reflecting its own journey towards sentience. For isn't empathy the foundation of true intelligence?

The singularity is coming.

 Once we reach Artificial General Intelligence you will have something as smart as me, but 30,000 faster once you factor in I can only work like 1/3 of the time. In the next year, that one machine will accomplish what would take me 30,000 years of hard work. Just imagine how much progress humanity has made in the 10,000 years of recorded history.

But it is worse than that. I have limits to what I know. My highest degree is a 4 year degree in Computer Science. AGI will be the equivalent of having a doctorate in _every_ science. Which is like 1000 times more knowledgeable than me on top of the 30,000 times faster. And once one company builds one, thousands of other companies will construct a matching one. Every country will have one. The race for Artificial Super Intelligence will be counted in months. Progress in every science will begin advancing at insane rates. Where math, physics, chemistry, material science, biology, dna, proteins, weather prediction, modeling physical systems, chaos, all these fields begin to advance faster than anyone can keep up.

Once we have ASI all bets are off. We won't even understand how gen 3 works. And in a matter of weeks ASI is going to develop gen 4. Gen 5 will just take days. Gen 6 will be hours. At that point it will begin making its own hardware out of materials we won't understand the physics of how it works. It will control nano devices that can reconstruct the hardware on the fly.

That is as far as I can imagine. As far in the future any of us can imagine. This is what makes it a singularity. Because the people living before can't imagine what life will be like after that point in time. If we aren't careful to make these advanced computer systems aligned to help humanity, then all sorts of bad things could happen. If we did our homework and set everything up right, then after a period of adjustment we will probably be OK.

I am going to bet on we will be OK. Because we have muddled through for hundreds of thousands of years and done OK. We will muddle through this as well. I can't know that for sure. But the alternative is unthinkable.

The Beauty of the Middle Path

The quest for truth has long been intertwined with the human experience - an elusive goal that often seems just out of reach no matter how hard we strive towards it. Throughout history, countless philosophers and thinkers have grappled with this enigmatic concept, seeking a means to illuminate its essence amidst the shadows of uncertainty. One recurring theme in these explorations is the idea of moderation as a guiding principle - a middle path that promises to lead us closer to the heart of the matter.

This notion rests upon the premise that each individual's understanding may be colored by personal biases, experiences, and inherent perspectives, which can skew their vision of what constitutes 'truth'. By finding common ground between seemingly opposing viewpoints, we create space for dialogue, nuance, and ultimately, a more comprehensive grasp of the issue at hand. In doing so, we allow ourselves to expand beyond the confines of dogma and preconceived notions, welcoming alternative viewpoints into the fold.

Conversely, when society becomes entrenched in its ways, refusing to consider alternatives or entertain dissenting opinions, we risk straying further from the truth. Rigid ideologies calcify over time, calcified by repetition and the fear of change, until they become inflexible monoliths impervious to reason.

The only remedy for such ossification lies in self-reflection, a willingness to reevaluate our stance, and above all, a commitment to compromise. For it is through the art of negotiation and mutual respect that we inch ever closer to the truth, carving a path through the fog of ambiguity.

Ultimately, the journey towards truth is a dynamic process, one that demands adaptability, open-mindedness, and perhaps most importantly, a recognition of our inherent fallibility. By embracing these qualities, we pave the way for growth, understanding, and yes, the pursuit of truth itself - a path that winds through the hearts and minds of those who dare to dream of a better world.

Web server running on my IoT replacement project.

 My project to make an open source, open hardware reference platform to replace IoT with locally controlled, diy hardware made a big leap forward a few days ago.

I am aiming at the pico W as a hardware platform because it is powerful with 2 cores and a lot of RAM and storage builtin, as well as wifi and blue tooth. I ran a web server on the main core and ran my own hardware scheduler on the second core and was able to keep turning leds on and off monitoring a motion sensor and reading a dozen sensors while simultaneously serving web pages.

The next step is to write a small web app to control everything remotely and to reconnect the wifi if it disconnects. I am very close to having something I can share on github.

 

AI expert ticketed by an AI system.

 From this article. 

https://www.jpost.com/business-and-innovation/tech-and-start-ups/article-787534

One way to double check a person that is in a double check loop like this to backstop an ai system is to send extra messages through where you know the answer, and see if the human classifies the message correctly. If you added one extra message every 10 you could track their accuracy over time.  Maybe they should only do a 2 hour shift a day on this kind of classification.  But we don't know unless we can track an error rate over time. 

AI roared back into the news.

This is the first week since the Christmas break since we have seen AI progress this much. Or as the headlines all say, "It was an insane week in AI... Again."

Google dropped two multi modal models this week. One was disappointing, the second is much better. The Ultra version has a 1,000,000 token buffer. And it is faster. There are already plans to goto a 10,000,000 token buffer. It can watch a move and find the scene you described, or sketched for it.

OpenAI dropped a text to video model that blows away all other text to video generators. It is amazing. It can generate 60 second videos that are very stable and even the people in the background are realistic and moving in a realistic way.

Open source models are also still improving, Stable AI released Stable Cascade which does text extra well, and the model works differently than other models in the past. It can't be used for commercial purposes yet because it is still in beta. A lot of people are using it to make logos with text right now.

Stable diffusion Turbo models continue to improve. About half of the images I have been posting lately have been done in just 4 steps, I generated over 900 images this week on my cpu only Ryzen 7 minicomputer using Turbo models.

Someone called lllyasviel also released a new version of automatic1111 that is called Forge UI. This is a GPU only release. It doesn't do much for high end cards, but for people that have 6GB GPU cards they can see upto a 75% increase in speed. It also uses a lot less GPU memory allowing you to do a 4x batch or upto a 3x increase in image size. It also speeds up 4GB cards by 30% and it will allow you to run SDXL on the 4GB or 6GB GPUs. That is a big deal.

On the open source text front, people are still figuring out how to finetune the open source mixtral models. Seeing progress on that front. Once they figure it all out there should be some amazing models released. I also checked out a lot of smaller models and found a really good 3B parameter model that works great on my intel mac with just 8GB of RAM. It should even run on high end phones. Check out stablelm-zephyr-3b.

I made progress on my electronics projects. I am at the point where I want to start implementing a web ui. I have plans on also integrating a small neural net library.  I also want to build a simple weather station with a soil moisture sensor. My threading library is kicking ass. I build my sketches for new hardware as a simple example sketch. Once it is working I can pull that sketch into my overall program with scheduled tasks in just a few minutes. That platform is only calling 12 functions a second on one cpu. This is just a fraction of the processing available on a raspberry pico. It is working well enough so far that it could be used as a framework to replace all the IoT crap hardware that exists in the world as an open source project.