Let’s talk about the difference between startups that will win using AI and companies that will lose using AI.
1) You will lose using AI if you’re simply improving upon current limitations. If part of your IP is extending context length or building out a serial networks to improve output, then the updates on the platforms will put you out of business.
2) If you are planning on LLMs replacing sales and support agents anytime soon, you will lose using AI.
3) You will lose using AI if you believe that platform tools like ChatGPT will do anything more than be a better version of their current selves.
Eventually, on a long enough timeline, I’m sure we’ll get Sam Altman’s stream of AGI. But it’s not happening anytime soon and too many people are making false promises.
So here’s how you win with AI.
1) If you are using it as a pattern recognition engine, whether it’s the platform where you’re building your own models, this is a great way to win with AI, and there are lots of problems that can be solved using this approach, but for the privacy of those working on those projects, I’m not going to be giving specific examples.
Here’s an example I’ve not seen, that makes the point. Imagine using machine learning to identify the patterns of marine life when they are in the presence of a sunken treasure. A properly trained model might lead you to right to the pot of gold.
2) You will win with AI if you use it as a convenience, not as a primary feature.
3) And you will win with AI if you use it based off what it is now, not on what’s promised.
If you’re using these tools every day like I am, you’re very familiar with its limitations. When you’re doing something easy, you have lots of competition. When you’re doing something hard, that’s valuable. That’s a great way to build something amazing.
