AI is here and it's staying for Good!
plus a short-poem about a hopeful future, written by Omni AI
To kick things off here’s a 5-line short poem, written by Omni of AlterAI.
A hopeful tomorrow
A hundred years from now,
I will live in a shack,
And there I will be laid to rest.
Now as I lay here in this bed,
I can see the future clear.
There have been way too many things going on. All along the same lines. I’m super excited. And I wanted to write this message for so long. It’s just that everything else got higher priority…
I’ll start with OpenAI Codex. This is mind-blowing-bending-twisting and all the adjectives out there. Codex is an algorithm that writes code. It’s based on GPT3.
I don’t think I could ever get bored testing it. And I’ve only spent so little time so far, most of the time using it was while doing Youtube videos on it (here, here, and here).
I think I’ll dedicate at least one full week to testing Codex. I’m not exaggerating. I asked it to write a game from scratch. And it did that flawlessly. I also asked it to code me some cybersecurity tools. And it didn’t fail.
I will use it for machine learning purposes as well. It not only writes but also explains code. I am simply bewildered beyond imagination. People don’t realize, but there’s humanity before Codex, and then there’s humanity after it.
Codex and nocode tools and platforms (like BubbleIO) are most beneficial to traditional coders and developers because they have the right mindset, algorithmic thinking, and the code knowledge to 10x their development process. And Codex and nocode enable that. For private lessons on how to become a developer using nocode, feel free to contact me on Twitter.
I will be making many more videos on Codex and how non-coders can build stuff (whatever you can think of) with it. Plus, I am integrating Codex into AlterAI.
Second, the hegemony of GPT3 is over. GPT3 is amazing, but the gatekeeping has to end…And it already has!
Folks at AI21 Studios have released a very capable AI model with 178 billion parameters, that seems to be better in some ways than GPT3. Rumors have it that part of its training was made on the PILE dataset. So, it might be smarter than GPT3 because the PILE dataset is of much higher quality than the data that GPT3 has been trained on.
My critique to folks at AI21 Studios is that they are very vague in their release paper for their 178B model.
That being said, anyone can get access for free to the Jurassic-178B AI model, right now. Sign-up via ai21.com
I might integrate Jurassic into AlterAI. I’m talking to folks at AI21, so we’ll see.
Third, on the personal side of things, I’ve been biking hundreds of kilometers in August so far, and I’m loving it. When my mind is racing with GPT3, Codex, and Jurassic thoughts, the bike is my relief valve.
Also, I’ve learned that B vitamins in the right formulation for bioavailability, alongside a healthy microbiome, are a maker or breaker of optimal physiology. I’m learning a lot about Akkermansia Muciniphila.
Fourth, the books that I am currently reading:
I’m gonna dedicate time to individually review The Polymath and for Designing the Mind because they are simply beyond valuable.
Feel free to reach me out on Twitter with thoughts about what I said here or to tell me what I can do to help you grow in any way, shape, or form.
Also, expect more frequent messages from me!!
Cheers, Cristi
If you want to get updates on stuff I’m doing, you can subscribe here: