Agreed. Six months ago, it was basically a gloried grammarly.
But lately, I’ve been taking hints from the “company brain” models, where it develops a running model of the case, and assesses each new piece as it comes in and updates the file.
I’ve also been using “Ralph Wiggum”-type models where you pass letter or contract drafts back and forth between agents with different goals (rules compliance, grammar, conciseness, ai slop detector, an opposing counsel critic, etc.). After a few rounds, it’s not perfect — but I start with a very good first draft in my hands.
This seems like a very similar set of tools to Anthropic. But I’ve enjoyed using the various frontier models to criticize each other.
Using recursive loops, the output has gone from a high school level intern to a 2nd year lawyer in about a year. It still doesn’t beat the experts, but so much legal work is (legally significant) pedantry, not legal philosophy.
AI will not kill off lawyers, or reduce the amount of litigation. It will increase volume and velocity.
I'd put my money on "yes," here. Of course, OAI can pay them off. But you know it had to happen to more than HuggingFace, and so somebody's going to try it.
They seem to be okay with an American slowdown, even though the Chinese won’t slow down. So maybe they want something the Chinese already have: liability protection.
The models spontaneously hack everything important without shame.
The frontier labs are going to get enjoined and regulated twelve ways to Sunday if the Feds don’t socialize the costs.
I've only heard of OpenAI and Anthropic agents hacking during training. Are open weights model not smart enough to hack Hugging Face to copy solutions? (doubtful)
On acquisition pricing: Why smarter AI models could drive up compute prices 10x (at least temporarily). [0]
This seems mind blowing, but the big boys seem to be behaving as if it's directionally true.
If compute is constrained and expensive, OpenRouter is what you'll use to get around the constraints at individual providers.
I had ChatGPT translate some old, handwritten French legal documents for family history purposes. It was far more accurate than I expected.
At scale, with better models, we might have a way to clear out the old archives. Not only could you translate, you could ask it to triage the discoveries. "Would the average person find this noteworthy?"
I have a ton of handwritten German stuff from the 19th century. My grandmother could make a fair stab at it, but nobody left can read it. I've shown modern Germans and they are at a loss. Thanks for your idea, I will give it a look. Any tips on model/method/training?
Try both Gemini Pro and ChatGPT. They are both outstanding at reading almost-unreadable documents. Use the highest thinking level your account supports.
(If you want to post a sample or two here, I'll try it. I like to collect difficult out-of-distribution test materials.)
1. "Conversation" is purely anthropomorphism. It's software input and output. If the client makes an excel spreadsheet about the cost benefit of ripping off people, it's not work-product.
But the lawyer's draft damages analysis in excel has always been protected.
2. If we're going to buy the "conversation" conceit, lawyers talking to consulting experts have always had a lot more work product protection than testifying experts.
The lawyer talking to Claude feels like talking to a consulting expert, especially since Claude can't have independent knowledge of facts that would allow it to testify.
A spreadsheet I produce for myself probably isn't attorney-client privileged. A spreadsheet or word document I produce for my lawyer can be attorney-client privileged (especially if it's literally only for the lawyer). If a ChatGPT or Claude chat is legally like a spreadsheet, it sounds like it's probably not privileged, but a ChatGPT chat you create "for your lawyer" would be.
I’d guess Lexis did that to itself. Usually the “deal” is that West or Lexis provides codification and reporting services for zero dollars, if they are named the official printer.
So the Legislature doesn’t have to maintain and oversee their own nest of troublesome legal pedants, and picks up a few contributions from legal publishing “entrepreneurs.”
By making the Annotated code official, it meant that anyone looking to prove what a particular law says in court would have to get it from the expensive, $412 hardback book, not the free version. I’d guess Lexis asked for that provision as part of its deal with Georgia.
Perhaps Sears/Kmart could have filled the Target niche, and leveraged Sears' brand portfolio rather than selling them off to make crushing interest payments.
But lately, I’ve been taking hints from the “company brain” models, where it develops a running model of the case, and assesses each new piece as it comes in and updates the file.
I’ve also been using “Ralph Wiggum”-type models where you pass letter or contract drafts back and forth between agents with different goals (rules compliance, grammar, conciseness, ai slop detector, an opposing counsel critic, etc.). After a few rounds, it’s not perfect — but I start with a very good first draft in my hands.
reply