1] It’s only because language models compute rather than think or feel that they’re unlikely to ever match humans in complex decision-making and knowledge creation.
2] The biggest achievement of LLMs is enabling the transfer of thoughts and ideas to and fro between our minds and databases.
3] AI will increase the net number of jobs in software because of the Jevons Paradox.
4] The users don’t care about what a marvel of engineering your product is; all they care about is whether it makes their lives 5x to 10x better or not.
5] However, people like watch connoisseurs keep praising the engineering behind their toys. This phenomenon only happens for physical products and not for software, as one can’t see or comprehend the backend.
6] Pragmatism in business increases the chances of success but decreases the probability of breakthrough success, because breakthroughs mostly belong to moonshot ideas.
7] If you’re building a software product that instantly excites the most non-technical people around you, then it’s probably not going to make a lot of money.
8] The B2C segment is largely composed of operations wrapped around software, for example, ride sharing, food delivery, and e-commerce. Tangible operations are the real moat, especially in the post-LLM world.
9] However, for a purely software B2C product, the only way to grow is to either (1) pour millions of dollars to drive behavior change, build distribution, and design (e.g. DuoLingo), or (2) to create something extraordinary and hope for virality (e.g. WhatsApp, Telegram). In other words, chances of success for a purely software based B2C product is very slim.
10] One neglected reason why software makes the world a better place is that it has significantly reduced the need for capital to merge value-creation with creativity.
11] As a technology, nuclear fusion will create a much bigger seismic shift than LLMs because energy is a more fundamental need (with only one source, the sun) than computation.
12] The most important innovations are those that directly address climate change and affordable cures for terminal diseases. Yet these receive limited focus, largely because their long revenue cycles don’t align well with typical market incentives.
