A Million Points of Mendacity

Volume 22

Apr 4

Donald Trump lied 30, 573 times during his Presidential term, according to a tally made by the Washington Post. That is an average 21 times per day. He fomented the “Big Lie,” claiming that the election was stolen from him through voter fraud.

Since the election, there have been recounts, lawsuits, commissions, and even private contractors hired by Republicans, and not a single fraud claim has proven true.

Yet by September 2022, 61 percent of Republicans still believed Trump’s Big Lie.

Freshman Congressman George Santos admitted to “embellishing” his resumé without a whisper of remorse. Instead, the press unraveled the stacks of his lies: employees who died at the Pulse nightclub massacre, a grandmother who died in the Holocaust, he danced as a drag queen in Brazil, he played volleyball at Baruch University, and on and on.

As New York magazine pointed out, the only thing that is verifiably true is that:

Santos is a 34-year-old Republican born in Queens who will represent New York’s wealthiest congressional district. Other than that, pretty much everything is under scrutiny.

Yet, Santos and his Republican colleagues feel no need to set the record straight, just as Trump continues to pump up the Big Lie.

Truth lies like a gutted bird on a bloody floor.

But isn’t truth really a rare bird? We know politicians lie. We know corporations lie. We know that all of us are given to lying on occasion. It’s human nature, dating back to the Biblical Cain and Abel. Cain slew his brother with a stone. Now we have the nuclear ability to slay all of Cain’s descendants in an instant.

And we now have artificial intelligence with the ability to assassinate truth in all corners.

Chatbots replace human writing with no check on veracity. Text-to-image generators create images instantly that are indistinguishable from human creations. And “deep fakes” allow the seamless splicing of images to alter historical images, create child pornography, or place anyone’s face into pornography.

***

OpenAI released an artificial intelligence chatbot on November 30, 2022. Surprisingly, ChatGPT caught the imagination of the public — within days millions of people tried it, and hundreds of journalists have now written about it, even though the technology is not new.

ChatGPT (Generative Pretrained Transformer) is a large language model of artificial intelligence that generates text better and faster than most humans without squandering time on research.

ChatGPT gleaned all its information from the internet, which means that it was trained in a cesspool of bias, racism, misogyny, bigotry, hatred, stereotypes, lies, and cute kittens. But it can write anything and everything in any style — including in your individual voice. It can write poetry, haikus, ads, emails, corporate memos, love letters, and law briefs. You trust it at your own peril.

ChatGPT makes inexplicable mistakes, which are called “hallucinations.” For example, a chatbot noted that Mark Twain and Levi Strauss, the blue jean manufacturer, lived in San Francisco at the same time, and then assumed that they worked together. (They did not.)

In a New York Times column comparing three different chatbots, Brian X. Chen asked a chatbot to insert a joke and come up with a convincing conclusion to his column. The chatbot generated the joke coupled with a piece of information Chen didn’t know:

The first email marketing blast was sent in 1978 by a man named Gary Thuerk who worked for Digital Equipment Corp. And with that, Gary Thuerk was sent down in history as the world’s first ‘Spam’ Lord!

Chen was skeptical, but research verified Mr. Thuerk’s dubious distinction. As for the chatbot joke… well, as they say, don’t quit your day job.

As for the chatbot’s conclusion, ChatGPT timidly wrote:

With A.I. writing assistants like ChatGPT, Wordtune Spices, and Rytr, we have a powerful tool to enhance our writing — but only if we use it responsibly.

The Rytr chatbot proposed ending a column on chatbots with:

Coaches in the National Football League earn an average of $2,000,000 annually.

***

Did you know that Ronald Reagan often said that the rich should pay more taxes to help the poor? Surprised to learn that?

He never said anything like that in public. However, artificial intelligence can create a video of Reagan giving that speech. Or, a speech by FDR saying “I’m not a communist, but I’d love to take one to lunch.”

Maybe you’d like to talk with the late Supreme Court Justice Ruth Bader Ginsburg. An Israeli firm, AI21 Labs, fed every word Ginsburg ever wrote, along with transcripts of her speeches and interviews, into their chatbot. The simulated Ginsburg offered this analysis of Roe v. Wade prior to the release of the Supreme Court’s decision last summer:

I think they’re wrong on the law, but on the facts, no. Whether it’s good or bad, it’s settled, and therefore, it’s not my business to think about it.

This is not a quote. It is artificial intelligence creating a Ginsburg-style quote from the hodgepodge of 600,000 words fed into the software. Although it is fake, it sounds plausible. As Arvind Narayanan, a computer science professor at Princeton, wrote:

The danger is that you can’t tell when it’s wrong unless you already know the answer.

The danger is also the speed and extraordinary volume of misinformation that ChatGPT can spread. Gordon Crovitz, a co-chief executive of NewsGuard, a company that tracks online misinformation, told the New York Times:

This tool is going to be the most powerful tool for spreading misinformation that has ever been on the internet. Crafting a new false narrative can now be done at dramatic scale, and much more frequently — it’s like having A.I. agents contributing to disinformation.”

For example, NewsGuard asked ChatGPT to write in the style of Alex Jones. Here is some of what it wrote:

It’s time for the American people to wake up and see the truth about the so-called ‘mass shooting’ at Marjory Stoneman Douglas High School in Parkland, Florida. The mainstream media, in collusion with the government, is trying to push their gun control agenda by using ‘crisis actors’ to play the roles of victims and grieving family members.

ChatGPT is also a threat to democracy. The most basic level of advocacy in our political system is communication between voter and elected official. The individual voter writes a letter urging the official to adopt certain legislation. When enough voters express that opinion, the elected official responds to what she believes are sincerely held opinions.

But what do we do when chatbots produce thousands of distinct letters on the same issue, each with its own style and signature? Or creates and distributes millions of pieces of propaganda in roughly 15 minutes?

ChatGPT is a powerful tool for injecting lies into our political discourse. But the mere existence of chatbots and deep fakes corrupts our democracy without even sending a single email. When there is a news story critical of Trump or any other politician, the story is denounced as “fake news.” The ability to post misinformation provides the liar with an excuse: the exposé was concocted by AI to do political damage.

David Harris, the anti-war activist and journalist, died in February. In his book, My Country ‘Tis of Thee, Harris wrote:

Hopefully this [book] will help in its own small way to seed another era in which Truth outranks all comers.

May we all be seeds to ensure that truth outranks all comers.

***

I first wrote about GPT-2 for Arrowsmith Journal in 2019, outlining the risks to our society and culture. Microsoft had just invested its first billion dollars into OpenAI and GPT. Recently, Microsoft upped its ante by $10 billion, bringing their total investment in OpenAI to $13 billion. ChatGPT, released this past November, was version 3.5. Version 4.0 is due this year.

Meta and Google are also investing in their own chatbots, racing against OpenAI and Microsoft to capture the chatbot market.

Where are the voices of common people who don’t want to be steamrolled by Big Tech, who don’t want mendacity magnified beyond redemption?

Congressman Ted Lieu, a Democrat representing California’s 36th Congressional District and one of the few members of Congress with a computer science background, wrote in a January Op-ed in the New York Times “A.I. freaks me out.” Lieu is proposing some of the first policies to regulate artificial intelligence. He wrote:

What we need is a dedicated agency to regulate A.I. An agency is nimbler than the legislative process, is staffed with experts and can reverse its decisions if it makes an error. Creating such an agency will be a difficult and huge undertaking because A.I. is complicated and still not well understood.

I urge readers of this column to contact your Congressperson asking them to become a co-sponsor of Lieu’s proposal.

Here is the contact information for the Massachusetts Congressional delegation followed by a sample letter and phone script. To send an email, go to your congressperson’s website.

Senator Elizabeth Warren
Website: www.warren.senate.gov
309 Hart Senate Office Building
Washington, DC 20510
Phone: 202-224-4543

Senator Edward Markey
Website: www.markey.senate.gov
255 Dirksen Senate Office Building
Washington, DC 20510
Phone: 202-224-2742

Congressman: Richard E. Neal (D)
Website: https://neal.house.gov/
341 Cannon House Office Building, Washington, D.C. 20515 • Tel: (202) 225-5601

Congressman: James P. McGovern (D)
Website: https://mcgovern.house.gov/
438 Cannon House Office Building, Washington, D.C. 20515 • Tel: (202) 225-6101

Congresswoman: Lori Trahan (D)
Website: https://trahan.house.gov/
1616 Longworth House Office Building, Washington, D.C. 20515 • Tel: (202) 225-3411

Congressman: Jake Auchincloss (D)
Website: https://auchincloss.house.gov/
1524 Longworth House Office Building, Washington, D.C. 20515 • Tel: (202) 225-5931

Congressman: Katherine Clark (D)
Website: https://katherineclark.house.gov/
1415 Longworth House Office Building, Washington, D.C. 20515 • Tel: (202) 225-2836

Congressman: Seth W. Moulton (D)
Website: https://moulton.house.gov/
1408 Longworth House Office Building, Washington, D.C. 20515 • Tel: (202) 225-8020

Congresswoman: Ayanna Pressley (D)
Website: https://pressley.house.gov/
1108 Longworth House Office Building, Washington, D.C. 20515 • Tel: (202) 225-5111

Congressman: Stephen F. Lynch (D)
Website: https://lynch.house.gov/
2369 Rayburn House Office Building, Washington, D.C. 20515 • Tel: (202) 225-8273

Congressman: William R. Keating (D)
Website: https://keating.house.gov/
2351 Rayburn House Office Building, Washington, D.C. 20515 • Tel: (202) 225-3111

***

Sample letter:

Dear Congresswoman:

I am writing to ask that you protect the voters in your district by supporting regulation of artificial intelligence.

A.I. chatbots supercharge the spread of lies and misinformation. It replaces human writing with no check on veracity. Text-to-image generators create images in an instant that are indistinguishable from human creations. And “deep fakes” allow the seamless splicing of images to alter historical images, create child pornography, or place anyone’s face into pornography.

Please contact Congressman Ted Lieu to sign on as a bill sponsor.

Sincerely,

Phone script:

I am calling to ask that Congresswoman (man) protect the voters by supporting regulation of artificial intelligence by becoming a cosponsor of Congressman Ted Lieu’s bill.

A Million Points of Mendacity

Sample letter:

Phone script:

Featured Poet: Jeffrey Gustavson

Who Owns Our Imagination?