"Wallet-size recorders that use A.I. to log live interactions have
become a product category.... A.I.-generated transcripts... preserve all sorts of things — offhand comments, quickly corrected statements, jokes — that humans would rarely write in the meeting minutes.... In a lawsuit or an investigation, that can make every word uttered discoverable. Even worse, say corporate lawyers: Sharing the meeting with an A.I. bot may void attorney-client privilege.... [Another] concern is accuracy. An A.I. transcript could, for example, record 'does matter' as 'doesn’t matter.' If that sentence comes up in court years later, the mistake may be difficult to remember. Corporate lawyers also worry about A.I. note takers’ lack of context and discretion. For example, recording every word of a board meeting, no matter how tangential the remark, could be legally perilous...."
From
"All Those A.I. Note Takers? They’re Making Lawyers Very Nervous. A trendy productivity hack, A.I. note takers are capturing every joke and offhand comment in many meetings. They could also potentially waive attorney-client privilege" (NYT).
Human memory is also not perfect. Who remembers old conversations and can testify about them with verbatim accuracy? And yet we've relied on frail humans all these years.
Posts
34 comments:
I thought the point of note taking is to be present so you're not just taking notes, you are also listening to all the information.
Sending an AI note taker ... to a meeting where the presenter used AI to generate the slide deck and their own remarks.
How beauteous mankind is! O brave new world,
That has such people in't.
It's just lawyers trying to figure out how to make money off of the AI infusion into their profession before it completely takes over their jobs.
So if your input is not required, why is it called a "meeting"?
AI note takers have the same problem of humans in hearing something and confusing it with something else. Whether it is homophones or a dialect pronunciation that isn’t familiar. Add in crosstalk, and the AI notes are confusing. An actual recording would have been better.
People tend to trust machines more than humans when handling personal information, often believing that machines are more secure and trustworthy. This is referred to as "belief in the machine heuristic" and this is going to be a problem when AI starts fact checking.
I watched the Christopher Nolan Odyssey trailer yesterday and was surprised at how manipulative and devious it was. Nolan praised his movie's ultra diverse cast, a Black Helen of Troy (actually she was from Sparta), a Black Athena and so on, and lectured us how great it was. But not one Black face in the trailer! Seriously, if you love diversity so much, own it. Anyways, I asked my AI some questions about this creep and his movie and was traumatized. I was told I was a racist and that Helen of Troy never existed. She was a fictional character. And guess who else was fictional? Jesus! The records of his crucifixion were all forgeries and the contemporaneous accounts of his life were all lies. That was not enough. If Jesus did exist, he was Black and, on second thought, Helen of Troy actually existed and she was also Black and a ton of Ethiopian soldiers were trying to rescue her from some sort of Greek rapists (this time they were White men). Also, Aesop, the famous Greek fabulist, was Ethiopian. The notion that Aesop was Black was sort of a gag started in the 18th century by William Godwin. What is AI's problem? Why does it have to sound like a mentally ill college coed.
AI is going to be far more accurate and "fair" when making decisions.
When everyone is represented by an AI lawyer everyone will be treated by the system equally.
I cannot wait until human lawyers and giant rich law firms are completely replaced.
Our judicial system is made up of people who overwhelmingly hate our country and have no sense of Objective Morality and therefor come up with complete bullshit like Roe v Wade and NFIB v Sebelius.
imTay said...
So if your input is not required, why is it called a "meeting"?
These are usually HR driven episodes.
The manager needs to make sure everyone "heard" them so they can cover their ass later when things fail.
"The difference between a good administrator and a bad one is about 5 heartbeats."
Bad administrators have a lot of meetings and people to blame. The worse they get the more people they have to blame. Eventually government has so many people who fucked up you can't actually find out who screwed it all up.
Good writing is good thinking: something AI cannot yet do.
AI is brilliant at giving your writing a veneer of intelligence. But it won't make your writing interesting if you have nothing interesting to say.
An AI transcript is unlikely to come into evidence in a lawsuit without a live witness who can testify to its accuracy.
steve uhr said...
An AI transcript is unlikely to come into evidence in a lawsuit without a live witness who can testify to its accuracy.
You mean like radar speed detectors and red light cameras?
There will be very "interesting" discovery proceedings in the near future...
I feel sorry for AI! It is criticized for its badness without mentioning the comparative badness of the human. Yes, there's this possible spark of creativity and judgment and life to the work of the human....
There’s seems to be a fundamental misunderstanding about “AI”. It’s not intelligent. It’s not doing analysis. It’s simply generating text based on how likely words are to be found together. So it looks like something you’d see from an expert, minus the knowledge, analysis, and accuracy.
As Theodor Roethke said to an unpromising poetry student, in his own account, "In the meantime, avoid all language."
I like to wear around my neck my thumb-size personal AI note-taker assistant. I can take a long morning walk along the beach spewing brief thoughts orally into audio files. When I return I merge those files and AI assembles for me an organized report of my random thoughts with AI commentary. It’s amazing how much otherwise lost content you can capture in just a week.
Most people recognize that humans are fallible, while experts say AI is not. But, evidence is proving otherwise. Also, humans can be cross examined, while AI cannot. Isn’t there something about the accused being able to confront their accusers?
Great! Can we now expect that even in-person meetings will be totally scripted? What a waste of time.
AI agents add most value to people who know how to ask the right questions and spot an incomplete answer.
The legal profession needs to move away from low turnover/high billing work to high turnover/much lower billing work, where the value added is to use AI to work out more quickly what the client might want to do and then to implement that in as automated a way as possible.
Some lawyers with good foresight will build really profitable law firms doing this. Most law firms will have to accept much lower profits from trying to avoid change.
It would be more efficient to send a kilowatt AI at minimum wage.
I admit that in my little group one of those note takers asks to be admitted to the (Teams) meeting and we always deny it.
A human taking notes can always ask for clarification. I certainly wouldn't want to depend for clarity on the same system that captions YouTube videos.
Shortly before I retired from practicing law, there was a lot of discussion about replacing court reporters with AI. The consensus was that it would never work. Misspelled words and proper names, different dialects, nuances of meaning, and worst of all -- people talking over one another. A court reporter can stop a trial or hearing at any time in order to clarify and insure the accuracy of the recorded proceedings. Maybe AI can do this some day in the future, but we are still quite a long ways from that.
“In a lawsuit or an investigation, that can make every word uttered discoverable. Even worse, say corporate lawyers: blah, blah, blah….
Hee. Lawyers know that AI is coming for their jobs. Judges, too. The sooner, the better, say I.
I wish I’d had one of these things for some of my college classes. Most of them, now that I think about it.
I sit on a few boards, and I can attest that while humans make transcription mistakes and errors of recall, so do A.I. "note takers".
There is a reason that most guides to parliamentary procedure insist that minutes are the only definitive records of meetings and that minutes only should include the basics, e.g., the time the meetings started, who was presiding over the meeting, who attended the meeting, the specific votes that were carried out, any executive decisions that were within the presiding officers' discretion, any announcements or committee reports that were requested to be attached to the meetings, and the time the meeting was adjourned. No commentary. No details of the discussions.
On several occasions, I have insisted that A.I. note takers be disabled, and I also have withdrawn from meetings when they were not disabled.
Zoom meetings used to start with some personal chatter, as a way for the attendees to connect with others who they might not see in person. During the meeting, somebody might make a joke. Or, in the course of discussing their work, someone might bring up a problem with another employee who isn't present. Now, you can expect every word to show up in print -- or, worse, every remark summarized in the AI's own language -- and emailed to what could be a long list of people who don't have the perspective of actually having been in the meeting. Right now it's probably causing a lot of embarrassment -- and in the future, as people get used to constant AI surveillance, it will become stifling.
Maybe the concerned corporate lawyers could just, I don't know, record the audio of the meetings and preserve the recordings?
I wonder how many years of AI taking notes of AI-presented "meetings" it will take before corporate management realizes what everyone else knows-- that "meetings" are generally a waste of time.
If something were important, I would never trust AI to summarize it for me. When I got out of the hospital, I would open letters from the medical team or lab results and that little box would pop up at the bottom of the screen asking if I wanted an AI summary of the document. Hell no! There are things in life that you really need to do yourself. (And if AI is doing all your important stuff for you, is AI actually living an important part of your life for you, too?)
lonejustice--I don't know where you practiced, but in CT the "certified" court reporters successfully excluded "transcriptionists". It was a jobs thing.
Never in my many years of practice did a "court reporter" ever stop the proceedings. All they did was record--audio record-- what was stated "on the record"--and if a judge ordered it, or a litigant filed a formal request, or an appeal was filed they would "transcribe" the recording--at a generous rate per word.
Well, I spent my entire legal career practicing in Iowa, and believe me when I tell you, the court reporter was the most important person in the courtroom outside of the judge. And she would even stop the judge if the judge was speaking too fast, or if a judge and an attorney were talking over each other. It didn't happen very often, but since what she transcribed would be the official record of the proceedings, and would form the record for any appeals, it had to be painstakingly accurate.
By the way, a court reporter doesn't transcribe on anything like a keyboard or typewriter. That would be way too slow. She uses a special machine with a much abbreviated shorthand type of entry. But it's the only official record of the trial or hearing, and she is the one who has to transcribe her record into a readable form to the judge and all the attorneys if requested. I was even taught in law school to respect her as the official keeper of the record. And if she asked you to stop so she could clarify the record, then you stopped. Same with the judge.
Charlie Curry: Drug sniffing dogs and traffic cameras can’t be cross-examined either.
Post a Comment
Please use the comments forum to respond to the post. Don't fight with each other. Be substantive... or interesting... or funny. Comments should go up immediately... unless you're commenting on a post older than 4 days. Then you have to wait for us to moderate you through. It's also possible to get shunted into spam by the machine. We try to keep an eye on that and release the miscaught good stuff. We do delete some comments, but not for viewpoint... for bad faith.