#OpenAI’s #GPT and its many related capabilities like #ChatGPT will never replace well trained, experienced human analysts. But I am experimenting with ways to make these tools more relevant to human analysts. Would love your feedback. Pls try out: unrestrictedintelligence.com
Conversation
Replying to
Cool idea! Results were sadly disappointing.
I used the prompt, “How should power grid operators prepare for cyber attacks?” The output was repetitive pablum and misinformation worse than no advice.
I give it a 10/10 - Great cyber weapon to unleash on adversaries!
1
1
Replying to
Thanks for that! Not ready for prime time for sure. Please come back and run that same query when GPT-4 comes out (but I might try now to improve some of the guidance I wrap queries in when prompts are submitted).
1
1
Show replies
Replying to
How recent is the dataset? Same as #ChatGPT ('21)?B/c I asked "assess most likely COAs for Russia in ukraine in 2023," & it began with "Russia is most likely to continue its current strategy of maintaining a presence in Ukraine through proxy forces and political influence..."
1
1
Replying to
Yep, huge weakness is all training data is prior to Sep 2021. Really looking forward to GPT-4 to see how current it can be.
2
Replying to
A colleague who works in the IC has been using #UnrestrictedIntelligence to assess the likelihood that the Russian Federation might invade #Ukraine. It was trained in 2021 and has no knowledge of the invasion. So far, it does not look promising. Output below:
1
Unfortunately the GPT-3 models were trained on data no more recent than Sep 2021. The world changed since then! Please keep kicking the tires and asking for intel assessments, will help us train the system, but big improvements not expected till GPT-4
Show additional replies, including those that may contain offensive content
Show