Dario Amodei of Claude said. Even if the accuracy of AI increases from 90% to 95%, the average user will not notice the 5% difference, he said.
But in the B2B world, that 5% is tremendous. In manufacturing, it improves yield and boosts profit margins. Even in coding, if you can do it 5% faster and with 5% fewer bugs, That is a level of evolution that humans can already experience. In other words - the improvement in AI accuracy happens for businesses, not for the average consumer.
In fact, in the course of creating B2B AI SaaS myself, I have seen time and again that the accuracy of that 5% is directly linked to economic returns. So the race to develop AI will definitely concentrate on B2B in the future. SMBs and individuals will not reach that level of difference. Countless AI vendors will crowd around large corporations. Capital will flow there, development will heat up and competition will accelerate.
But this is fate. Markets that have been proven valuable by AI, be most quickly price-destroyed by AI itself. The number of competitors increases quickly, All companies offer similar precision, similar prices and similar terms. Differentials disappear and profit margins continue to fall.
In time, the "how quickly can we replace humans with AI" will be the will be the sole axis of competition. Engineers will be under the illusion of increased demand, The content of that demand is supported by AI making AI. The moment supply leaves human hands, human bodies and minds will not be able to keep up with its speed.
That's where my concern lies. By accumulating efficiency gains of 5%, 5% of the human margin is being cut by 5%. The more AI evolves, the thinner the engineer's psyche is stretched, Eventually, it will become transparent.
What is the fate of AI products? Every time we increase the accuracy, we are slowly chipping away at the human mind. What remains after that, perfect deliverables and a creator that no one remembers.