Everyone in EU affairs knows that the Draghi report on the EU's competitiveness has just been released.
It's a 400-page detailed analysis that most people have no time to read.
Those experimenting with AI/GPT have already fed it into ChatGPT and asked a summary...but there is so much more you can do with it:
1) Given the sheer size of the report, a simple upload and prompt to summarize it often yields mediocre or somewhat superficial results. A somewhat better way is to start the prompt by saying "Take your time to carefully analyze this document in depth and then come up with a summary that focuses on [...] in a [...] format". This often improves the results.
2) You can typically get even better results if you cut a large (30+ page) document into 15-20 page segments and then request various summaries for each 'small' PDF, or upload each file separately and instruct the GPT to "look" at each one by one, thereby avoiding "cognitive" overload and potential omissions. (This can also be done as a chain-of-thought, ie. step-by-step prompt.) You can use the free Smallpdf.com for this purpose, and they also have an AI-powered chat for PDFs that you can upload to their platform (like Adobe Acrobat's recently integrated paid service).
3) A more advanced level of working with complex reports of this kind is instructing the AI/GPT to focus on direct and indirect messages in the report that are relevant to *your* sector or industry, instead of requesting a generic summary. For example, if you work in the solar panel sector, use a prompt that says "Pay special attention to topics and issues that would be highly relevant, even indirectly, for the solar panel industry".
4) You can also say "Generate 10 points from the report that could be the basis for a press release that our [...NGO/trade association/company...], working in [... sector/area/policy..] could use for a press release to highlight issues that are important for us." - you use this after the above-mentioned summary, or base it on the original raw report (or PDFs).
5) You may also experiment with the above summary/extract and feeding in your organization's (public) annual report or strategic plan or policy paper, and asking GPT to "analyze how the Draghi report's findings compare to [our] plans, and where are the weak/strong points in the latter".
The possibilities are endless, only limited by our own creativity.
Hope this post gives you a few more ideas, though ✌️
p.s.: if you're interested in learning more about AI/GPT for public affairs, advocacy, and EU policy, join one of our 6 *free* webinars starting on 23 Sept 2024 with Jacques: https://lnkd.in/e6npR854
**This was originally posted on Andras Baneth's LinkedIn account.