Sometimes I wonder whether we can really prompt our way out of this, particularly in these cases where it’s not a super common scenario and requires actual critical thinking and not just emulating responses to similar common scenarios, without just providing such granular instructions that it kind of defeats the purpose. Of course, if we could just make a version of Copilot with custom prompts specific to our organization’s audit procedures, then it would work. I just mean you need so much expertise to even prompt these things, it doesn’t seem like general intelligence is even possible yet.
Really good observation - it’s entirely possible that GenAI, at least in its current state, might not be able to consistently or reliably do this type of assessment. But the best way to determine that is through larger amounts of statistical testing, I think. Worth a shot right??
![[88] The AI Accounting Games - may the smartest model win!](http://i.ytimg.com/vi/BkL6JHBQJNo/mqdefault.jpg)
![ลองสุ่มไอโฟน 990 โกงมั้ย? [ โกงมั้ยครับ ep.101 ] | DOM](http://i.ytimg.com/vi/bfsWe0j492k/mqdefault.jpg)
Sometimes I wonder whether we can really prompt our way out of this, particularly in these cases where it’s not a super common scenario and requires actual critical thinking and not just emulating responses to similar common scenarios, without just providing such granular instructions that it kind of defeats the purpose. Of course, if we could just make a version of Copilot with custom prompts specific to our organization’s audit procedures, then it would work. I just mean you need so much expertise to even prompt these things, it doesn’t seem like general intelligence is even possible yet.
Really good observation - it’s entirely possible that GenAI, at least in its current state, might not be able to consistently or reliably do this type of assessment. But the best way to determine that is through larger amounts of statistical testing, I think. Worth a shot right??