Greetings, everyone!
Today I participated in a Tech Conference for Developers of all spectrums, and it is worth mentioning that although the conference had ten panels running in parallel for different disciplines of software development, Generative AI was independently present in every discussion. A huge spectrum of tools and ecosystems was exemplified in various talks, from code generation, testing, documentation, bug reporting, architecture design, and even talking about complete product development and the role of the software engineer in the post-GenAI world.
On the way back home, triggered by some discussions related to the intrinsic bias built into the Large Language Models, I ran a small experiment, and I asked various AI Assistants to give me a number between 1 and 25. I gave this request to Gemini, Chat-GPT, Claude, and Grok with the expectation that I would receive a broad answer, as one would expect from even a basic form of intelligence.
Are you surprised to learn that the answer received was the same from all four AI Assistants? Yes, 17!
Different AI assistants, trained with different data and running slightly different models, with different parameters, concluded that 17 is the most desired number.
I don't know about you, but I find this disturbing... Diversity will be reduced to a single, standardized truth?