81 Comments
User's avatar
Robert's avatar

Weirdly though, the chatbots are still 50 times more accurate than Fox News, and 50,000 times more accurate than a white house press briefing.

Expand full comment
Mike McCabe's avatar

Major accuracy issues across the four AI chatbots studied ranged from 18-22%.

Karoline Leavitt "Pffft, that's nothing."

Expand full comment
Greg Kapphahn's avatar

I suspect Lyin' Levitt does her best to shoot for 100% INaccuracy each and every day. I wonder if she can order off a restaurant menu without lying? When she travels, how can she get where she wants to go when the truth simply can't escape her mouth?

Expand full comment
Robert's avatar

Ha ha.

"I will never lie to you," she lied, sweetly.

Expand full comment
Michael Baker's avatar

LMAO

Expand full comment
Adam's avatar

Ahahahahahahahahaha! Damn right Robert.

Expand full comment
BTAM Master's avatar

Eventually that may not be true: AI "learns" by reading the Internet...and the more bogus info AI reads from the White House and Fox and Friends, the more it will repeat it, read it again...and you can see where this is going.

Expand full comment
Julie's avatar

Bahahahaha so true

Expand full comment
B Dodson's avatar

But AI will use "Fox News" and the "White House" as news sources, so what could go wrong?

Expand full comment
Frank Lee's avatar

Did you hear that from CNN or The View?

Expand full comment
Robert's avatar

No, direct from Fox and Leavitt.

“The president supports the right of Americans to peacefully protest,” she said. “He supports the First Amendment.” One of many Leavitt lies.

And Fox? That trump beat Biden in 2020? That's the whopper, but they lie daily. They paid out a billion dollars for their lies about voting machines. They lie about immigrants, democrats, January 6th, trans people... everything really.

Expand full comment
Greg Kapphahn's avatar

Of course the weasel lies. As everyone knows, actual, factual truth has a liberal bias.

Expand full comment
Therese S.'s avatar

Thanks for the review, I don't plan on getting that AI browser. Actually, I hadn't planned on it, but now I definitely won't. I notice AI keeps popping up all over the place, and it slows me down because I have to take time to swat it away. <frown> Very annoying!

Expand full comment
Adam's avatar

"I notice AI keeps popping up all over the place" Ok? I don't need help articulating my thoughts, don't want AI attempting to usurp my place.

Expand full comment
Judith Swink (CA)'s avatar

And now we have: I just received this FB post from Meta!

"What you should know

On December 16, 2025, we're making changes to our privacy policy. Here are some details.

Personalizing your experiences

We’ll start using your interactions with AIs to personalize your experiences and ads.

What this means

Personalizing your experiences includes suggesting content like posts that you may find interesting and reels to watch. It also includes showing ads that are more relevant to you."

Expand full comment
Adam's avatar

Oh boy! You mean they're gonna tell me what I like and I save time and money if like what they say right away? Sweet! Sign me up MetaMan! What's not to like?

Expand full comment
Judith Swink (CA)'s avatar

Exactly. And I've received a second FB post from Meta about this new AI assistant but haven't looked at it yet.

Expand full comment
Adam's avatar

Why look? Save time! You already know you're gonna like it, right? I mean, duh... why else would they send it to you, right? I mean, like right? Right?

Expand full comment
Judith Swink (CA)'s avatar

Just came across this WaPo article (gift link). It just gets 'better and better', no?

https://wapo.st/4npxbe9 "ChatGPT just came out with its own web browser. Use it with caution."

Expand full comment
Charlie Cooper's avatar

Two more problems with chatbots:

They use the content of real news providers, but divert the public from actually landing on the providers' websites, thus undermining revenue needed to do actual fact-finding and reporting.

Second, they use all that energy and water, the cost of which is often dumped onto the public.

Expand full comment
JenneJ's avatar

The cost of AI and bitcoin generation in energy and water consumption should make both those technologies cost prohibitive.

Expand full comment
Mike McCabe's avatar

No wonder this Administration is so pro-AI.

Expand full comment
Larry Carr (autocarr)'s avatar

Baron has taught pops to use AI —now able to drop shit on America with ease. Here comes 1️⃣💩

Expand full comment
Carole Tarrant's avatar

“Hal, don’t make shit up.”

Expand full comment
Joseph Mangano's avatar

I wonder what goes into the thinking behind these AI interfaces. Do the programmers/executives pulling the strings believe that a non-response makes their product look ineffective?

The defense for a lot of AI's flaws is that the technology is still in its infancy, but if it's going to keep pulling from the same polluted pool of misinformation, I don't know how it will meaningfully improve. Garbage in, garbage out, as they say.

Expand full comment
QOTM31's avatar

LLMs don’t actually know anything, and they are designed to give you whatever it thinks is the most likely next chunk of words based on its probability modeling. They will virtually always give you some type of answer, because they’re designed to. It is possible to steer LLMs to limit responses, or not respond in certain contexts, but it takes training and is generally done for specific purpose models. News and current events are not a good use of this tech as that training data constantly changes and facts evolve, which LLMs have no context for.

Large AI companies want to feed the perception that LLMs are magic all knowing oracles in service of their growth and making money. Hallucinations are inherent to the technology. They are a tool, useful for some things, not so much others, and they have limitations. In my personal opinion, if this tech was so amazing and capable, companies would not be shoving it into everything and trying to force people to use it.

Expand full comment
Judith Swink (CA)'s avatar

"Large AI companies want to feed the perception that LLMs are magic all knowing oracles in service of their growth ..."

Much like the way so many have perceived the internet itself, as the source of all information and knowledge, with no idea of how much of what they find is not always accurate, or even complete.

Expand full comment
NubbyShober's avatar

AI-tweaked news will thus be a huge boon to RW media outlets like FOX News, that depend almost entirely on fabricated or massively distorted information; and who could care less about inaccuracy in reporting, as long as it helps the GOP.

Expand full comment
Katy Bolger's avatar

I'm not sure why we need AI news when we have the news. What is the advantage of asking AI a question over looking it up? This is a former humanities teacher's worst nightmares and I got out of teaching high school about two minutes after we heard the phrase AI and I have not regretted for a moment not needing to chase down an increasing number of false and made up sources. Yuck. What a fn time and energy waster AI is.

Expand full comment
mark's avatar

Artificial flavorings, Artificial limbs, and Artificial teeth (correctly named false teeth). These all correctly illustrate what artificial truly is in those examples and in AI. It is an inferior replacement for the real thing. AI is an inferior replacement for real intelligence.

Expand full comment
Gnomon Pillar's avatar

My rule of thumb with most salesmen (here we speak of Altman, Musk, Thiel, Andreessen, et al: when the hype and pitch for their product begins to increase exponentially, assume it probably ain't a good product. And it certainly won't perform as promised.

Expand full comment
Hannah's avatar

AI is not ready for primetime as a player.

Expand full comment
BTAM Master's avatar

As a retired IT pro who still programs, I found AI very helpful when working with some completely new technology: it could write the initial code for me and then I learned all about it as I Googled to fix the horror show it's created. Using it less and less every day.

AI learns by reading the Internet...and the more bogus info AI creates, the more it will read itself, meaning it can never improve.

Expand full comment
Robert's avatar

Judd, Will this information be broadcasted on the evening news, morning news, any mainstream news…hmmm…bet not! Great reporting on a growing topic that has real consequences. Keep the light shining! Great responses from all too!

Expand full comment
A Sarcastic Prophet's avatar

Oh the convenience to let somebody (or something else) think for me. But there are so many wrongs here in AI land. Where does one begin? Garbage in, garbage out. Hallucinations. Lies. 20% failure rate. Not to mention the environmental costs. And of course the ultimate extermination of humanity if not from an AI consciousness, at least from the ultimate destruction of our climate making earth uninhabitable. Happy Thursday indeed. Sometimes I hate you Judd for telling the truth. It hurts to know that I am complicit in my own destruction.

Expand full comment
JenneJ's avatar

I wish they had explained more about what exactly an AI hallucination is. Sounds fascinating.

Expand full comment
BTAM Master's avatar

An AI hallucination is when it gives you totally random unrelated information.

Nerdy example:

Me: "write code to do {blah blah blah} in java"

AI: "Here you go"

Me: "That's from an unrelated project we worked on last month and it's COBOL not java"

AI: "You're right! Good Catch!"

etc.

Expand full comment
JenneJ's avatar

Thank you BTAM Master. That is helpful. May I inquire further?

At that point does AI correct the mistake and give you what you asked for, or could you get another hallucination? When you ask your AI for the blah blah blah java code next month does it correctly give you what you ask for, or could you get another hallucination? Is it learning?

Expand full comment
BTAM Master's avatar

Usually...but you might (or might not) get another unrelated surprise.

Sometimes.

Yes, but since it can't tell the difference between good source and bogus source, you never know what it's learning.

Try asking chatgpt "Draw a Where's Waldo picture" and see what you get.

Good luck!

If I could really answer these questions, I'd be a very successful stockbroker with a crystal ball.

Expand full comment
JenneJ's avatar

Thank you.

I asked ChatGPT to write a sci-fi, sarcastic, dark, dramedy, screen play titled "The Real Lives of AI Wives". It was hilarious.

Expand full comment
BTAM Master's avatar

Try asking for a romantic poem...

Expand full comment
Linda’s Morning's avatar

At some point would you please investigate the resource costs of AI?

Expand full comment
Therese S.'s avatar

I think it's already been shown to waste electricity and water.

Expand full comment
Linda’s Morning's avatar

Absolutely, but the story should be spreading far and wide…people should know that the water use for one transaction comes from their daily use of water. Data centers will be using up people’s water resources. It’s really part of the big picture in my mind.

Expand full comment
Jean(Muriel)'s avatar

So, will we, as usual hear, see, be told that this is not an accurate way to get truth and still continue down the AI rabbit hole?

Get the word out that REGULATION is freedom not to be lied to. It is freedom in knowing your education is not hammered out by thugs who have one thing in life to do: that is to control your freedom. Resist! Demand! Be smart! Make America Honest!

Expand full comment
Greg Kapphahn's avatar

Trouble is, honesty cuts into profits, and truth has a liberal bias to "conservatives" whose entire perspective is build on lies and denial of the truth. The AI folk will NOT let truth and honesty get in the way of profits. They just hope to invisibly squirrel away massive amounts of investor money before the bubble pops.

Expand full comment
Randy Dyck's avatar

2% mistakes make most news outlets unreadable. Thus why would I trust chatbots. AI has a way to go.

Expand full comment
JenneJ's avatar

I don't know where they got that statistic, but do you believe it? Seems to me there are a lot of news outlets out there make way more than 2% "mistakes" and loads of people are still paying attention to them.

Expand full comment