Step into the future of AI
Today, I had an excellent day, while I spend my day with learning. I watched several videos regarding the actual trends of AI.
After watching these videos from high level to code level, I'm more convinced about my next vision is not so far, and who can leverage those nation states will be superior.
While major companies fighting for enormous GPU capacity:
There is another path. Especially with roots of Hungary among so much talented people such as John von Neumann or Ede Teller. If we extend the scope to the region we can also name Nicola Tesla.
What is common with our roots? As we didn't have enormous resources, we always have to come up with a smart solution which can be done with the limited resources.
What happened when they reached the land of promise? They show brilliant performance.
The same such I designed in the past an alternative to an Oracle Engineering System for 20% of the cost 7x performance - NOT BAD!
This is not different with the field of Artificial Intelligence.
Nation states race against each other to have supremacy over the competitors. BUT!
They use brute-force methods to have results!
All the existing methods use exponentially complex methods, which I understand in case of solutions such bitcoin where energy and computation capacity describe the power of wealth, but in case of Artificial Intelligence it is a waist.
领英推荐
Because of the time pressure in this race, companies use the old mindset - you have more money than time - Burn it Baby!
However if you think smart you can use other fields methods what already exist - and this is the old recipe which always work.
After today's lectures I'm convinced how we can overcome the resource constrains what the current brute-force method creates.
With some layering, and use of field intelligence, there is a way to decrease the requirements with 100x, compared to the current resource needs.
I'm sure the results will be the same if it doesn't better.
If I get once dedicated time, to work on these, I will define the game-changer AI strategy, that build even more powerful models for special purpose.
The world is going into these directions, just to name some concepts which using the same patterns:
I just imagine what happens if a nation state who have the actual resources start building such AI models.
There is also a logical order for the development - and I think the focus must be on the coding model as a start.
If you have the coding model, it will assist you on building the next models framework. This way you can multiply the programmer resources, and compete even with the largest tech companies.