Related Posts
Thoughts on Micromass?
Hi, I was a Teacher by profession for 8 years and last 1 year I worked as AI Product Associate. I basically write content for python, Data Science and Artificial intelligence with some hands on projects. I want to grow and find more opportunities. Can anyone suggest what is best for me? My current package is above 7 lakh.
Newco Tata Consultancy Accenture ZS Associates Infosys
What is the pay bracket for DI AM level (A&A)?
Happy Beach - Bench Day

Additional Posts in Law
New to Fishbowl?
unlock all discussions on Fishbowl.



Junior lawyer. Westlaw Deep Research is a helpful first step for finding helpful cases, running it in parallel with boolean searches and reading secondary sources. I’m still a better writer than AI and don’t see myself using it for drafting. Despite the sales pitches it never really does find that “needle in the haystack” in production document sets.
I don’t see it changing the skills i feel i need to develop, it might just be another tool.
I think the big problem with AI is you can’t trust it for document review.
If you’re somewhere that the details of the record are important, it’s gonna be a liability because it will miss things.
I would check out Ross Guberman’s stuff on AI. He has a lot of insight on what AI can do that’s helpful when you’re in a position that you can’t let it do your work for you.
Thank you for the reference. I agree with you, ai can miss things so we need to check outputs.
F
I have no idea where AI is going. We have tried a couple AI programs on a trial basis. Both were only okay. I would say they were decent at giving you a general idea of what the records say, but we obviously have to then review the AI output. AI just seems like somewhat of a time saver. That's it. We are starting another trial with a third company which starts with a T which is giving an extremely hard sell and wants total control over the trial to the point where I question it, especially given how the other vendors gave free access. They are allowing work on one case only per person during the trial. And the output is not a few minutes after the prompt is given. It's a couple hours. What?