Where does AI lose?
Examining Rough Edges
AR has more potential than AI. Its flaws are easier problems to solve. And their solutions are headed in the right direction. The problems with Large Language Models are possibly fundamental issues. These being,
Hallucinations
Trust is a fundamental quality of software. Without it, even products with the biggest potential are doomed. If we can’t trust that the model will give us a correct answer or at least take a less certain tone. ChatGPT has a disclaimer that they may give incorrect information
Claude has no such warning.
Yet often its answers give no indication of the possible incorrect information.
False Promises
No code tools are promising to build entire applications. They don’t address fundamental issues
What if there’s something the AI can’t do?
What about adding on features beyond launch that are too complex for the AI to implement?
How to educate the founder on what’s actually going on in the code so that they’re not completely blind when they need to fix something or make an adjustment?
Market Saturation
The AI boom is more reminiscent of the dot com boom than the mobile app wave. So many companies seem to be founded on a wing and a prayer. Or worse yet, are squeezing AI into the product for no particular reason other than having AI in the product.
Prediction
AR will outpace AI.
The cost and convenience benefits of reducing people’s entire hardware life down to one device are larger than flawed efficiency improvements.