Jaws dropped this month when Open AI demoed their text-video generator Sora. It generates high-quality, smooth, realistic videos of whatever crazy shit you can think of. People who know about these things were amazed at the leap in quality from the already impressive video-generating models. The nerds discuss the physics involved and speculate how this fantastic tech works.
It's easy to opine on the cultural implications of realistic video generation. If this works the way some think it does, it suggests that we now have a way of generating the training data to train robots that can operate in the physical world. In other words, this fun toy is a better, faster, cheaper way to train human-like robots.
A big deal from Google isn't the rebranding of Bard to Gemini; it's the expansion of the context window. That's the nerdy way of saying how much junk you can dump in for the AI to reference. Gemini can handle up to 1 million tokens, which is way more than the others. It's an hour-long video, entire books, and many PDFs. The race will continue pushing all the AIs to get faster, cheaper, and able to handle larger volumes of information. Now, what does it mean to be an expert? Let's imagine a bright intern with an AI that knows almost everything about the business and all the other stuff the AIs know. Or an offshore employee? What does this do to the value of a knowledge worker?
Knowledge retrieval, reasoning, and prediction are going to be done by AI for a fraction of the cost of meat computers. That leaves judgment and goal-setting for human brains along with the coordination and management of the AIs. This implies a profound shift in the organization of corporate labor. Some predict that a billion-dollar company could soon operate with a staff that can share a pizza.
Ethan Mollick's blog and social posts are worth following. This article, "Signs and Portents," explores this topic more deeply.
Here is another article on how AI will change jobs.
https://briancheong.medium.com/how-will-ai-change-our-jobs-ddab439616ab
AI Manifestos
Philosophy was a minor subject in most people's education. I bet you don't know any philosophy majors. But that's the skill set we need to decide how to use AI and create policies and procedures for employees, customers, and other stakeholders.
Writing policies is complicated because the tech is so complex and changing fast. Some companies instead create manifestos that outline their vision and values and offer a perspective.
I'm creating a consulting product and writing a book to guide organizations in crafting manifestos and policies because I think creative thinking could enhance this task.
Midjourney
I've had access to the Midjourney Alpha website. It's way easier to use than the geeky Discord interface. The quality of images V6 generates is phenomenal. I'm scratching my head, imagining why anyone would spend tens of thousands of dollars a year sending their kid to an art school.
Midjourney plans to add social features soon, and I just learned they are working with X on a partnership. They are also working on some hardware devices. I can only imagine what kind of crazy shit could come from a combination of a camera and Midjourney AI generations.
Apple Vision Pro
I have yet to have a chance to try Apple's new VR goggles. But for the first time, my mimetic desire is piqued. I even stopped to look at Meta's VR stuff while passing through Target the other day. Perhaps VR is about to take off. Will this mean I can take the ugly TV off my wall? The unintended consequences of massive VR adoption and the normalization of wearing AI glasses will be fascinating.
AI Marble Sculptures
I’m going to Italy next week, and I’ve been reading about Michelangelo. It’s been fun creating my own marbles.