People often talk about “AI” in absolute terms, as if it’s a single entity:
“AI is great!”
“AI is the future”
“AI is terrible!”
“AI should be banned!”But AI is definitely not a single thing. AI is based on a broad range of technologies and applications, from machine learning algorithms to neural networks and beyond.
AI is a structured data set organized through training: Chat GPT, Gemini, etc., are trained on large (and different) datasets to recognize patterns and make predictions.
Differences in LLMs and algorithms: Each LLM and algorithm will be designed and implemented differently, with varying architectures, training data, and optimization techniques.
Bias in source data and implementation: AI bias can be introduced in the data, the training, or the algorithms. Biases in any of these areas will almost assuredly generate biased AI outputs.It’s important to approach discussions about AI with these details in mind. You wouldn’t generalize the entire human race based on a few individuals. Sweeping statements on AI are equally invalid. Each AI implementation has its own strengths, weaknesses, and potential biases. By recognizing this, we can have more informed and productive conversations about the role of AI in our world. Let’s strive for a balanced perspective! 
Alt: An image of a woman with ChatGPT and Gemini placed side by side with the caption: corporate needs you to find the differences between this picture and this picture. stating that they are all the same.
TL; DR: All AIs are different, and you can’t generalize them! 