There is a lot of coverage of how Michael Cohen found fictional cases on #Google Bard to give his lawyers, who then (apparently without double-checking with conventional sources to verify that they existed) gave them to the judge.
Of course this isn't the first time this has happened, but there's a much deeper issue here.
Most ordinary folks do NOT UNDERSTAND what AI and Large Language Models are about. They read AI company disclaimers that the firms know they are basically just trying to use to protect themselves -- not the users.
In particular, the world has been trained for decades to trust Google as a source of good information. Having Google-branded services that mix accurate and totally fictitious information into well written and convincing toxic responses is devaluing Google's core brand. This is very sad.
@lauren > "In particular, the world has been trained for decades to trust Google as a source of good information. Having Google-branded services that mix accurate and totally fictitious information into well written and convincing toxic responses is devaluing Google's core brand. This is very sad."
I mean, YouTube has existed for *a while* now.
At some point, the user has to accept responsibility for trying to drive a screw with a hammer.
@LouisIngenthron Blaming the user is always the wrong side of the argument. YouTube has never been purported to be any kind of authoritative anything. Google Search has been, and the generative AI add-ons are diluting it into uselessness. People have busy lives. They don't have the time or in most cases the background to understand this stuff even if they did have the time. Techies who try blame users in these cases are most of the problem.
@lauren I get that perspective, and it's totally valid, but there's got to be a limit. If a user writes their resume in Microsoft Paint, that's got to be their own fault.
@LouisIngenthron That's not what we're talking about. Hyperbolic examples are worthless. We're talking about people who go to Google to get accurate information, and how Google has confused reality and fiction by deploying LLM systems that are not ready for safe public use by any stretch of the imagination.
@lauren *That* I wholeheartedly agree with. The tech is not ready for mass deployment in established search systems.
But were Cohen's mistakes just provided to him via search, or did he actively seek out an LLM to do his work for him?
@LouisIngenthron What from what I've seen on this story, he went to Bard because he thought it was Google's new "supercharged" search system. Better than regular Search. In fact, that's a very common misconception.
@lauren Just tried it myself.
First, you have to agree to the terms, a two page document that, on the second page, has this bit:
"Bard is an experimental technology and may sometimes give inaccurate or inappropriate information that doesn’t represent Google’s views.
Don’t rely on Bard’s responses as medical, legal, financial, or other professional advice."
That should be moved up to the first page IMO, but right after you agree to that, a popup comes up that says right at the top:
"Bard is an experiment
As you try Bard, please remember:
Bard will not always get it right
Bard may give inaccurate or offensive responses. When in doubt, use the Google button to double-check Bard's responses."
Finally, under every prompt input field, there is the following text: "Bard may display inaccurate info, including about people, so double-check its responses."
And we're talking about an attorney who ignored or clicked through all three of these clearly-worded warnings and still didn't even think to double-check the results? At some point, the user does have to take responsibility.
We don't what the signup process and warnings were at the time Cohen used it. And one argue that google like every other company stuffs so much crap into the TOS and other legal boilerplate that users have become habituated to ignore it.
@artemesia @lauren Don't read the warning label on a power saw? Lose your hand.
Don't read the warning label on an LLM? Lose your law license.
I'm good with both of those. People need to be responsible for their own actions.
