re: Anthropic just studied which jobs AI can be theoretically replaced.

quote:

---

Member when posters ridiculed me for saying lawyers would be in danger of replacement?

---

No doubt there will certain tasks for which AI beats human cognition, but until you an sue a bot for malpractice, lawyers aren't going anywhere.

P.S. I guarantee you'll never be able to sue a bot for malpractice. The lawyers who write the AI terms of use will run the meter dreaming up new ways to disclaim liability for outputs. There will always be the need for a human to sign, represent, verify legal conclusions, and argue/negotiate.

quote:

---

Sonnet 4.6 is available in Anthropic's free tier, so money wasn't necessarily wasted to make a point.

---

Then your point, whether you intended it to be this or not, is you get what you pay for.

In your case, someone paid zero and got a stupid answer.

But what would be satisfactory one-word answers to the questions that were asked?

quote:

---

Then your point, whether you intended it to be this or not, is you get what you pay for.

In your case, someone paid zero and got a stupid answer.

---

The Sonnet 4.6 LLM is available on both free and paid tiers. I don't have insight into whether "Taya" paid to get a stupid answer. But I guess I could ask Claude to give me a one word answer on which tier she used

quote:

---

Everybody diving headlong into using it to take over the world need to pump the brakes. I've had a team of 6 very qualified people (AI is what they do) working for months on agents and all kinds of other stuff to try to perform what would have taken a moderately intelligent intern about a week to do. So far, nothing close to a solution.

---

AI is not at a point where it's a 1-1 replacement for a person. It will just take pieces of people's job's until it doesn't make financial sense to keep them, or certainly as many of them.

People would be shocked to see the number of hours out of 40 they spend on simple manual tasks that current AI can take over. All it needs is the access and a small amount of human input.

quote:

---

AI company says it can replace 80% of bosses.

---

Then I'll retire early. What do you think I can expect in severance pay?

I’m a lawyer, and I’ve spent a lot of my time the past 2-3 months incorporating AI into my practice. I’m a little over 15 years removed from law school. Here are my observations:

1) Engineering prompts / delegating to AI takes more time than delegating to a senior associate. However, if you give AI a detailed roadmap, it can chop off a ton of associate / review time. I’d say that a complicated motion to dismiss on an issue I did not have much familiarity with took 5-6 hours of my time instead of 2-3 hours of my time plus 20+ hours of associate time.
2) Litigators are safer than transactional attorneys. I think transactional attorneys will lose a lot of their workflow. As long as there are human judges, opposing counsel, ethical rules, human clients, etc., litigators will be needed. Business lawyers? Tougher sledding.
3) Due diligence and doc review hours will be gone. Paying an attorney to review thousands of contracts to flag certain key contract terms is over.
4) Young attorneys are in trouble. I’d rethink going to law school right now.
5) The work that you get from AI is serviceable but it is not inspired. It’s like B, B+ work. That’s probably fine for the vast majority of legal work.
6) This could be the technology that makes the legal profession work closer to the way it should work. The cost of litigation is obscene right now. I’d rather be in court and in depositions than behind a desk slogging through discovery.
7) This technology will come for everyone. Legal is impacted because the same code that applies to software engineers also applies to legal (ie they are both rules based).

In short, I’m glad I’m more than 15 years into my career and have some measure of experience and judgment. I imagine I will be doing more high level, judgment-based work, and will just be working with AI instead of young attorneys. Figuring out how to monetize the practice of law is going to be challenging.

quote:

---

I drew an extremely rudimentary floor plan onsite of our duplex.
Nearly illegible writing/numbering and loose doodles for the sink, appliances, tub, vanity, etc. Really just for me to understand as I was planning to transfer the info to a real floor plan ( I used to be an architectural draftsman).
The wife (no pics) got bored during the ride home and snapped a pic of my drawing and ran it through ChatGPT with a few statements regarding the property.
Fricking ChatGPT asked questions to verify the location of the rooms and then spit out a very comprehensive drawing in less than 15 seconds.
This was from a cell phone in the middle of the desert on I-40.

---

Maybe I'll use ChatGPT.

Because what was presented to me in no way made my job any easier.

Their AI really only worked on architectural plans - counting items that you told it to count (how many doors labeled Door A, how many windows labeled Window B, etc.)

But I couldn't show the AI a set of structural drawings and have it do a takeoff of concrete members like pile caps or grade beams. I think it could give me slab area, but wouldn't calculate cubic yardage based on slab thickness. I couldn't even get it to do a count of piles because the piles weren't notated, just a circle on the plan view drawing. That's next to useless for my line of work

quote:

---

This is where people are going to get trapped. Most people's experience with AI are annoying and borderline useless nerfed models. They get frustrated with how inept it is, and think it will never replace them. They fail to realize the acceleration that is taking place behind the scenes. These things are improving every day, and they will soon reach an exponential growth phase. Buckle up.

---

Is it really a trap if its as unavoidable as you say? Not like we can really do anything about it, if true.

quote:

---

Autonomous mowers that operate like Roombas are becoming increasingly popular

---

But can it arrange and design shrubberies?

quote:

---

Is it really a trap if its as unavoidable as you say? Not like we can really do anything about it, if true.

---

The trap is rejecting AI and treating it as a threat instead of embracing it, learning it, and implementing it in your life. For the forseeable future, and maybe always, there will need to be human oversight/cowork with AI. If you flat out reject it, you'll be one of the first on the firing line when (not if) it comes for your industry.