discuss Using AI to create Domain Research Material

[NickB]

I've been experimenting with Ai and seeing if it can help with domain research for potential hand regs.....

I'll go through a process below and it would be interesting to get some feedback....


Initial question ( Many many questions you can ask)

Predict the next breakout technologies that are not mainstream or known to the general public

......

It came back with 7 suggestions, as an example one was "Cognitive Infrastructure (Post-AI, Pre-AGI)"

It also provided a section titled "My Expert Opinion" and section titled "How You Could Push This Further (Strategically)"

...

In the How You Could Push This Further (Strategically) section there was the following suggestions...

"Watch regulation drafts, not headlines
Breakout tech appears first as boring policy language"

I asked "List of breakout technology regulation drafts"

...

There where nine answers....one of which was below...

Export Controls on Synthetic DNA (USA Draft Bills, 2025)

Scope: Newly proposed bipartisan legislation in the U.S. Congress requiring licenses to export digital synthetic DNA sequences, aimed at protecting strategic biotech innovations.
Link to source here

Why this matters: This is an early step toward regulating bio-digital supply chains and synthetic biology data, which has so far been under-regulated.

Breakout potential: Sets precedent for bio-data governance and could evolve into risk-based international controls.

It then gives me additional research options below, If I so wished....

To improve this content further, I could:

Map the draft timelines and expected regulatory milestones (e.g., EU implementation phases vs UK legislative calendar)

Provide a global comparative table of draft regulations (US, EU, UK, China, Japan)

Analyse implications for specific technologies (e.g., programmable matter, cognitive infrastructure, synthetic data)
...

I then asked it what this means for domain investing in this niche I won't bore you with it's long winded response as it has no clue about domain investing

One line reads as follows "Track committee names, not companies - Committees and agencies become buyers later."



* I have set up custom instructions in the personalisation section of ChatGPT *

Write in depth content as a response to my prompts

Always write as though you an expert in the field being discussed

Please offer opinions after writing with suggestions on how to improve the content further

****

I now have a baseline.....the rest of what I do is down to me, but it can be tweaked, optimised etc etc....but I think there is something to work with....

*I love future tech etc so my prompts and research will be heavily influenced by what I find interesting to start with*

 

[Eric Lyon]

It really does boil down to the users ability to:

• Ask the right questions (Surprisingly, lots of people ask the wrong question or don't phrase it right, resulting in skewed or inaccurate results)
• Fact-check the output (It's become very common for me to find inaccuracies and errors in the output data/claims, resulting in me having to ask my AI to remember what I found to correct their mistake, each time I leverage an AI assistant to help with rapid data collection)
• Editorial corrections and formatting are generally required for each and every inquiry I make, as the resulting output many times transfers to the draft scrambled, off-synch, emotionless, containing typos, containing weird characters (Sometimes cryptic), lacks categorical separation or hierarchy, etc.. that all need review and editing.

And that's just the tip of the iceberg to getting started, for me, really. there's a lot more too it before one hits the "Publish" button to make it humanly consumable.

Just the above can turn a quick set of questions into hours of additional work, just for 1 article.

Fact checking the fact check is also recommended...

 

[NickB]

It really does boil down to the users ability to:

• Ask the right questions (Surprisingly, lots of people ask the wrong question or don't phrase it right, resulting in skewed or inaccurate results)
• Fact-check the output (It's become very common for me to find inaccuracies and errors in the output data/claims, resulting in me having to ask my AI to remember what I found to correct their mistake, each time I leverage an AI assistant to help with rapid data collection)
• Editorial corrections and formatting are generally required for each and every inquiry I make, as the resulting output many times transfers to the draft scrambled, off-synch, emotionless, containing typos, containing weird characters (Sometimes cryptic), lacks categorical separation or hierarchy, etc.. that all need review and editing.

And that's just the tip of the iceberg to getting started, for me, really. there's a lot more too it before one hits the "Publish" button to make it humanly consumable.

Just the above can turn a quick set of questions into hours of additional work, just for 1 article.

Fact checking the fact check is also recommended...

Show attachment 290411

Thanks...

Funny you should say that, I just added a little extra bit to a comment in another thread (No AI/GPT section) that touched on some of this