[WSBAPT] Off-Topic, but not Off-Putting

[Dave Culbertson]

So I guess to some folks my post was, in fact, Off-Putting. Oh well, can't please everyone.

I'm skeptical of AI myself, and don't think it is a substitute for human thought and action.

BUT: it's in development and is going to keep improving and changing and I would be surprised if it doesn't have some applications that don't currently seem apparent. So, these folks are working on that long-term process and I hope if anybody has some law firms that might participate in the project they'll throw some names in the hat.

Thanks.

dc

From: wsbapt-bounces at lists.wsbarppt.com <wsbapt-bounces at lists.wsbarppt.com> On Behalf Of Eric Nelsen
Sent: Monday, April 29, 2024 1:16 PM
To: WSBA Probate & Trust Listserv <wsbapt at lists.wsbarppt.com>; solo-and-small-practice-section at list.wsba.org; creditor-debtor-section at list.wsba.org; wsbarp at lists.wsbarppt.com
Cc: Business Law Section <business-law-section at list.wsba.org>
Subject: Re: [WSBAPT] Off-Topic, but not Off-Putting

I have nothing against the ActiveWords folks-I am responding here because the current AI hype is a huge problem, and I try to take opportunities where I find them to warn people off.
First, I highly recommend searching for UW computational linguistics professor Emily M. Bender, and her "AI Hype" postings on social media, YouTube, and elsewhere. She has testified in Olympia and elsewhere, and of course published professionally, on the hype and the social dangers of incorporating so-called "AI" into computers and systems doing substantive work.
https://time.com/collection/time100-ai/6308275/emily-m-bender/
Important points:
"Artificial intelligence" is a marketing term and is not a coherent concept or a particular technology.
The current AI hype in the legal field is largely about large language model (LLM) systems like ChatGPT that, in response to an input, generate text. The generated text is determined on a statistical basis derived from the contents of the database of text scraped by the creators of the LLM.
LLM responses are inherently untrustworthy because they are text intended to look like a plausible response. They are no more than that. It is basic to the structure and design of LLMs that they will generate plausible-looking but false responses a nonzero number of times to a question. This is not a correctable problem because the design of LLMs is founded on this statistical structure. Prof. Bender has called LLMs "stochastic parrots" because they imitate speech on a randomized (but statistically nuanced) basis.
When an LLM generates a response that looks like a "true" answer to a question, it only means that the statistical process generated a plausible-looking answer that we, on the basis of other information, and identify as true. If the exact same question is input a second time, the LLM will generate another plausible-looking answer that may or may not be as "true" as the other answer.
Thanks for taking the time to read this...

Sincerely,

Eric

Eric C. Nelsen
Sayre Law Offices, PLLC
1417 31st Ave South
Seattle WA 98144-3909
206-625-0092
eric at sayrelawoffices.com<mailto:eric at sayrelawoffices.com>

From: wsbapt-bounces at lists.wsbarppt.com<mailto:wsbapt-bounces at lists.wsbarppt.com> <wsbapt-bounces at lists.wsbarppt.com<mailto:wsbapt-bounces at lists.wsbarppt.com>> On Behalf Of Dave Culbertson
Sent: Monday, April 29, 2024 12:26 PM
To: solo-and-small-practice-section at list.wsba.org<mailto:solo-and-small-practice-section at list.wsba.org>; creditor-debtor-section at list.wsba.org<mailto:creditor-debtor-section at list.wsba.org>; wsbapt at lists.wsbarppt.com<mailto:wsbapt at lists.wsbarppt.com>; wsbarp at lists.wsbarppt.com<mailto:wsbarp at lists.wsbarppt.com>
Cc: Business Law Section <business-law-section at list.wsba.org<mailto:business-law-section at list.wsba.org>>
Subject: [WSBAPT] Off-Topic, but not Off-Putting

Hi, Folks.
I use a software called ActiveWords. It's not exactly what this query is about, but in case you are curious it allows you to type a couple of keys of your choosing anywhere you are working on your computer and it will go find and pop open a folder, file, website, or program instantaneously. It saves me a bunch of time and I highly recommend it. Happy to talk about it more if anybody's interested.
BUT that's not the immediate point. The founder of ActiveWords (they have spectacular customer support, BTW-that's how I know "the founder") is working on another project for lawfirms, and asked me this question below. So I'm seeing if the Hive-Mind has some answers:
___________________________________

[From Buzz Bruggeman, AW]

I have been advising/mentoring an AI legal tech startup here in Seattle.  They have some very, very clever and powerful technology.

Key focus right now is on firms that have/do...

a.              Doing primarily motor vehicle accident PI work.
b.              10+ lawyers
c.             30 new cases a month


Do you know say the names of firms that match the criteria above? Maybe firms you refer cases to?
Trying to use Google to search for firms that even tell you the number of lawyers in a firm is impossible.
Thanks in advance for any suggestions.

Best Regards,

Dave Culbertson

The Law Office of Davisson Culbertson
PO 20403
Seattle, WA 98102

Phone: (206) 478-8134
FAX: (866) 867-7796
dculbertson at culbertsonlawoffice.com<mailto:dculbertson at culbertsonlawoffice.com>


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.fsr.com/pipermail/wsbapt/attachments/20240429/1cbb5676/attachment.html>