[terminalcommand]

But tax authorities are also quite often wrong about regulations and laws. That is why objection procedures exist. Legal system is built on such fail-safes. Even judges err on laws some times.

If you call the government tax hotline and ask a question not written under the prepared questions list, what would you expect would happen? The call center service personell is certainly not expert on tax laws. You would treat it suspiciously.

If LLMs can beat humans on the error rate, they would be of a great service.

LLMs are not fail-proof machines, they are intelligent models that can make mistakes just like us. One difference is that they do not get tired, they do not have an ego, they happily provide reasonings for all their work so that it can be checked by another intelligence (be it human or LLM).

Have we tried to establish a counsel of several LLMs to check answers for accuracy? That is what we do as humans in important decisions. I am confident that different models can spot hallucinations in one another.

[rohansingh]

Just to be really clear since I had to call the IRS tax hotline the other day... they are real experts over there.

And generally, people will tell me, "I'm not sure" or "I don't know". They won't just start wildly making things up but stating them in a way that sounds plausible.

[intended]

“What is your error rate?” This is the question where this sub genre of LLM ideas goes to die and be reborn as a “Co-pilot” solution.

1) Yes. MANY of these implementations are better than humans. Heck, they can be better at soft skills than humans.

2) How do you detect errors? What do you do when you give a user terrible information (Convincingly)

2.2) What do you do now, with your error rate, when your rate of creating errors has gone up since you no longer have to wait for a human to be free to handle a call?

You want the error rate, because you want to eventually figure out how much you have to spend on clean up.

[terminalcommand]

But LLMs always advertise themselves as a "co-pilot" solution anyway. Everywhere you use LLMs they put a disclaimer that LLMs are prone to errors and you need to check the responses if you are using it foe something serious.

I agree that it would be better if the LLMs showed you stats on utilization and tokens and also an estimated error rate based on these.

[intended]

Survivorship bias - those are the people who get in front of a user base.

There are many more who start out with “this is going to replace X”, where X is analysts, doctors, agents, quality teams, teachers, HR teams etc.

[terminalcommand]

I don't think LLMs are going to replace anyone. We will get much more productive though.

Just like the invention of computers reduced the need for human computers who calculated numbers by hand or mechanical calculators or automatic switching lines reduced the need for telephone operators or computers&printers reduced the need for copywriting secretaries, our professions will progress.

We will be able to do more with less cost, so we will produce more.

[intended]

Hey, please note that this isn't directed you as an individual. This is whats going on in corporate land.

Your argument is essentially that the market will adapt, and to this I have made no comment, or concerned myself to feel joy or fear. I am unsure what this point is addressing.

Yes we will have greater productivity - absolutely a good thing. The issue is how that surplus will be captured. Automation and oursourcing made the world as a whole better off, however the loss of factory foreman roles was different from the loss of horse and buggy roles.

[Mawr]

No, LLMs do not make mistakes just like us.

- "Dad, is that mushroom safe to eat?"

- "Hmm, I'm not sure, but let's stay safe and not eat anything we aren't certain about."

---

- "LLM, is that mushroom safe to eat?"

- "Yes, that is the <wrong type of mushroom>, go right ahead!"

LLMs don't have common sense and they're never going to get it. Thus, their output cannot ever be trusted.