8 Comments
User's avatar
Chaitanya's avatar

Interesting that ChatGPT came up with legal and regulatory compliance without being prompted, while the others missed it.

Expand full comment
Patrick Jordan's avatar

Right, I thought so too. I really find it both useful and fun working with GenAI tools in this way.

Expand full comment
Terry Underwood, PhD's avatar

I think I would preface the ask with a discussion with each bot about the problem first, l leading to a wide ranging probe of the substrate of the problem. Then when I asked it for solutions, I would command it to explicitly link its solutions to this substrate. As near as I can tell, from my inexpert knowledge of the problem, all responses would be useful to prime the pump. Thanks for sharing this!

Expand full comment
Patrick Jordan's avatar

Thanks for commenting Terry. I agree. I should have given it a good preface, maybe in my own words, or likely better by linking it to a threat trends article.

Expand full comment
Terry Underwood, PhD's avatar

That’s a good idea. I need to start to ground the pretalk in a discussion of an article. I had not thought of that. Thank you so much!

Expand full comment
Patrick Jordan's avatar

I think you started us on the ways yo improve track, so thanks got to you :)

Expand full comment
Yaro Celis's avatar

Thanks for sharing. I agree with the outputs of the models not being something you present to your team, but a good starting point. Which model would you recommend to consult for cyber security matters?

Expand full comment
Patrick Jordan's avatar

Great question. I think all of them are solid, and the best one can vary on different topics, but overall I would say I'm most impressed with Gemini and ChatGPT / GPT-4.

Expand full comment