This was fantastic. I've been describing a similar "anything-to-anything machine" idea as a new ground for communication. The tradeoff between communicating at scale and communicating with high fidelity is minimized when I can losslessly translate between modalities. My only caveat is that even in the third scenario what you would have is the bifurcation of remaining human tasks. No matter how good an AI is at recombining itself (at least with current techniques) it's not gonna solve Moravec's Paradox and do human intuitive tasks well. Similarly at the height of decision making actions aren't always about knowing what's best but having the right intuition. CEOs and Carpenters will always be with us.
I agree, with caveats. The current spectrum of AI don't know what the world is, beyond extrapolating from what it reads. To get there it needs embodiment, which (imo) is essential to future development.
Re the CEO point my suspicion is that even intuition will get commodtitised soon, esp because it's effectively black box pattern recognition, but the underlying process behind the intuition does not. We still will have to drive to fulfil whatever we desire.
When do you think we will transition to Era 2 and 3? What are the impacts on human jobs in these two cases?
We are some way away from Era 2. We may require new algorithms, the next level of computing, and more quality/quantity data (In case we run out of the data, it may be synthetic data or data created by other models and fed to these models). What do you think?
Simply brilliant and even visionary essay. No mistake in subscribing to this newsletter! I would love to read your thoughts someday on the possibilities of an AI implemented post-scarcity civilization a lá Iain Banks's Culture.
An example of where AI generated prompts result in awesome AI generated images:
https://twitter.com/GuyP/status/1598020781065527296?s=20&t=sODodeedRrIlUZtBcV-SqQ
That's cool
This was fantastic. I've been describing a similar "anything-to-anything machine" idea as a new ground for communication. The tradeoff between communicating at scale and communicating with high fidelity is minimized when I can losslessly translate between modalities. My only caveat is that even in the third scenario what you would have is the bifurcation of remaining human tasks. No matter how good an AI is at recombining itself (at least with current techniques) it's not gonna solve Moravec's Paradox and do human intuitive tasks well. Similarly at the height of decision making actions aren't always about knowing what's best but having the right intuition. CEOs and Carpenters will always be with us.
I agree, with caveats. The current spectrum of AI don't know what the world is, beyond extrapolating from what it reads. To get there it needs embodiment, which (imo) is essential to future development.
Re the CEO point my suspicion is that even intuition will get commodtitised soon, esp because it's effectively black box pattern recognition, but the underlying process behind the intuition does not. We still will have to drive to fulfil whatever we desire.
Agree on embodiment
Excellent Post!
When do you think we will transition to Era 2 and 3? What are the impacts on human jobs in these two cases?
We are some way away from Era 2. We may require new algorithms, the next level of computing, and more quality/quantity data (In case we run out of the data, it may be synthetic data or data created by other models and fed to these models). What do you think?
Simply brilliant and even visionary essay. No mistake in subscribing to this newsletter! I would love to read your thoughts someday on the possibilities of an AI implemented post-scarcity civilization a lá Iain Banks's Culture.
That's wonderful! It's super cool to see how it's helped creators honestly, and with the possibility of enabling a full workflow later on.