The Amazing AI Revolution

hasn’t yet started as seen with errors in ChatGPT (gen-AI)

Popularity isn’t the measure of success in technology. Instead, the ability to use the technology is what matters to weed out errors in usability. Photo by Erik Mclean on Unsplash

Generative AI / LLMs like ChatGPT and Copilot have their problems: unnatural context handling, ‘hallucinations’ or making stuff up and lack of direct integration with Legacy applications.

Generative AI is claimed to be a revolution, but years after the revolution started, the biggest winners are the makers of benchmarks, not products because benchmarks continue to be improved on while no obvious improvement against human-level interactions register.

In my example below, the reality exposes how gen-AI misses the fact that human language is continually learned by people. Language isn’t something we understand everything of, but it is something we have a base capability. We can use that capability to learn new knowledge.

This is important as LLMs provide answers that are incomprehensible to untrained people and the mechanism to progress requires a lot of effort. Who you’re talking to is a key aspect of context that is excluded from today’s generative AI systems. It is needed to make the language comprehensible.

My Problem: Add a user to Azure

My requirement was for a team member to deploy a set of our production servers to a new platform. I signed up for the…