A reflection by Batiste Roger, CTO of Odonatech
There are... too many! To find my way around, I use this mnemonic trick that will certainly be useful for navigating generative AI: CAM³ERA.
What does CAM³ERA stand for?
It's just a checklist that allows you to assess what an AI is capable of:
C - Critical: Is there a control performed at the AI's output to validate its response?
A - Agentic: Can the system take actions, such as sending an email?
M - Multi-agent: Internally, are there multiple actors, like in a football team, or... in any company? Each has a specialized role, and unity makes strength.
M - Multimodal: Ability to process and generate various types of data (text, image, sound, etc.).
M - Multi-step: Can the system plan and execute tasks in multiple steps?
E - External: Can the system interact with external resources, for example by doing web searches or calling functions (e.g., interest rate calculator)?
R - Read/Write: Can the system read and write to one or more databases? Typically, I include RAG in this category.
A - Autonomy: Does the system have a certain degree of independence or even self-improvement? Can it (on its own) evaluate its performance and change its functioning based on past experiences?
What's it for?
With this checklist, when faced with a new AI that I'm discovering, I can:
Verify that I've understood what it does, in quite some depth, in an instant
Ask myself if it's bullshit or if there's a fair bit of intelligence inside (potentially we could have a great AI that only ticks one of the boxes, but an AI that ticks all of them is likely to be more sophisticated / powerful).
Personally, I find it really helpful.
Conclusion
As the field of generative AI continues to advance, tools like CAM³ERA become essential for navigating its complexity. Whether you're an AI professional, a business decision-maker, or simply a curious enthusiast, this acronym will help you better understand and evaluate generative AI systems.
Feel free to use and share CAM³ERA in your discussions about AI. Together, we can make the fascinating world of generative AI more accessible and understandable for everyone.
(It's obvious that an AI generated the conclusion, but the rest was good, wasn't it?)
Comentarios