Does it actually make sense for them to massively thin associate ranks if in law they're actual profit centers that bill hours? I've never looked to see if associates are actually profitable for the firm or if they're just loss leaders or net neutral and it's the partner hours that make the real money.

In finance juniors are almost purely a cost center. Making them redundant does nothing but raise margins. But it's at least moderately harder to fully automate a lot of finance stuff and the easiest stuff to automate are the parts of the job that are most law-like (interpreting transaction docs, writing).

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098241)

If properly managed, first years can absolutely be profitable, although it's not uncommon to lose a lot of money on them. In general law is like this; the difference between me making $950K and $150K is surprisingly thin and can come down to me being negligent in my billing and associate management, plus maybe one or two bad breaks on the cases.

AI is absolutely capable of replacing non-clerk first years, and I've written on here before that I quit using TTT SAs in 2023 (2022?) because of AI. Outside of top-tier biglaw/boutiques where HYS/A3clerk associates will continue to go, I'm not sure how the profession will bring up talent anymore. I'm sure it will, but it'll probably require a paradigm shift.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098252)

I would think that transactional law back and forth could mostly just be "Claude, make the indemnification language here less stringent and it shouldn't cover X,Y,Z"

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098292)

I'm in lit, where it's quite useful and specifically fills a role that overlaps heavily with what SAs and first years used to do: write research memos that sometimes slightly miss the point and almost never display particularly good judgment (personal, political, etc... you know the shit that judges actually decide cases on), but that get the partner up to speed on the law.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098328)

I find that AI is like having a super-diligent but naive junior associate working for me. I'll start doing my work and then assign the exact same thing to the AI and have it work on it in tandem. I'll be lazy and spend 1/3 to 1/2 the time that I used to when I was trying to do a really good job. Then I'll check the AI output against what I did.

A typical ratio for AI analysis is like: 30-40% misguided, hallucinated, or outright wrong, 45-55% correct and overlapping what I did, 15% genuinely clever or insightful stuff that I probably would have found or developed myself had I worked harder. I spend a bit of time developing that 15% and call it a day, and just cede whatever I could have found above and beyond the AI.

The bottom line is that it probably lets me be 95% effective while being lazy, whereas without it my lazy work would be 80-85%. I'm willing to risk the 5%, but being 15-20% off would be too much. Also, it's a lot more useful for me than it would be for some junior associate, because I can distinguish between the made up bullshit and genuinely insightful stuff on sight and ask the correct followup questions, whereas it would send juniors down numerous hours-long rabbit-trails.

The thing it really kills is contract reviews. Like if I have 10 different contracts affecting a case and they have some identical, some similar, and some different provisions, I can review one while it's comparing all of them to each other and figuring out what's what.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098316)

Add in a contingency practice and it's LJL (the associates know no one's paying but think that it's fulfilling their target hours). I've done fee petitions where I struggled to identify 100 hours of work product for 1,000 hours of billing.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098735)

How this impacts law firms remains to be seen. There could be a much higher deal volume because the transaction costs will be so much lower. The one thing that is clear is that no client is going to pay $900-1200 per hour for associates to do due diligence or document review. Anything that can be commoditized and done better by machines will be. What clients will pay for is judgment, anonymized knowledge about what peers/competitors/regulators/legislators are thinking, etc.

From a biglaw partner side, it will do away with a lot of the worst parts of the practice (at least from my view). But it'll make it even more important to be good with clients, to have good industry intelligence, etc.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098290)

You seem to assume partners do this out of necessity and not from innate sick sadistic pleasure they get from it.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098567)

"heh"

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098570)

1) the biglaw business model is almost entirely relationship based and AI doesn't move the needle at all on rainmaking as an operating function. There is no other product to sell besides billable work so associates aren't going away in commercial law but AI will definitely make litigation a race to the bottom and the outcome disparity between commercial/lit paths will grow bigger than ever.

2) unlike everyone else's clients in shitlaw or non-law businesses, it is suboptimal for any biglaw client to use AI to check their lawyer's work. It's objectively stupid as shit actually and only has downside. Whereas every dumb fuck shitlaw client who used grok for 20 mins is too stupid to realize don't do that.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098730)

2) Agreed. But no biglaw client is ever checking their lawyer's work. That's why they hire biglaw firms.

(http://www.autoadmit.com/thread.php?thread_id=5898136&forum_id=2,#50098742)