Koobdoo - Learning Equilibrium Mascot
The Learning Equilibrium

Wanted: 1 best-fit Generative AI platform?

By Sanjay MukherjeeSeptember 24, 2024

Pune, Maharashtra, India, 20.36 hours, 24 September 2024:

So, which AI platform does one go for? Is there one that fits best for all the work that I may want to do? Is there anyone out there who is likely to know?

There are hazaar (thousands) lists out on the World Wide Web and beyond. 'Top 5 AI', 'Top 10 AI', 'The Best AI Platform...' and so on. But they are pretty useless as a reference point for my work. Mainly because most influencers, evangelists and content writers write for a general purpose audience, not a professional audience (one that comprises people who specialise in a particular field).

This is a graphic I created in Apple Keynote and then enhanced with iColorama; For the life of me, I could not get any of the Generative AI platforms to auto-generate an image like this even with reference images. I have to work harder on my prompting, but that may not be the problem...

The graphic accompanying this little post is a list of the platforms that I have been trying out for the past few months. I will get around to writing my experience with these some time soon, meanwhile:

Psst, What I have learnt so far: There is no best-fit Generative AI platform. It really depends on what you want to do and in which field.

For instance, if I am looking to derive a requirement or build an iterative content structure or refine content through dialogue, I usually spend time with either ChatGPT, Stability, or Claude - sometimes with all three. If I am generating images, then I go to Leonardo or Runway or MidJourney or DALL.E or Stability ... depending on which field I am working on (e.g. food items, food ingredients, airline cabin crew training, restaurant food service, insurance sales agent work process, etc).

Learning: None of the platforms are good for all type of specialist image generations.

Generative AI is evolving very fast, but at the moment I have to train models if I want high fidelity, high accuracy contextual output for a specific field, which takes considerable time and money when compared to using a stock image or getting photos shot or getting a graphic artist to create a custom illustration.

But this gap is likely to be addressed fairly quickly.