🗞️ Chatbots deceived into spreading fake news

ALSO: A single ChatGPT user can replaces 50 workers

Estimated Read Time: 7 minutes

AI deepfakes are a danger to our knowing what’s real and what’s fake. But what happens when the AI itself can’t tell which news is real? That was the topic of a recent study by NewsGuard, in which 32% of the answers from big chatbots (like ChatGPT and Gemini) were false and cited fake news sites as their source. Let’s dive in!

• 🗣️ Chatbots can’t tell what’s real news.

• 🔔 The Pope raises the alarm about AI.

• 😵 A single ChatGPT user can replace 50 people.

• 💰 AI is changing how ads work.

• 🥊 Perplexity tries to make Google obsolete.

🗣️ Chatbots can’t tell what’s real news

What happened: Apparently top chatbots (like ChatGPT and Gemini) are spreading Russian disinformation. This is according to a study by news-monitoring service NewsGuard.

The details: 

• Leading chatbots (when asked about Russian news) would spread misinformation 32% of the time and cite incorrect/false news sites as a source.

• Chatbots repeated disinformation in 152 of 570 responses, sometimes adding disclaimers in 29 instances.

• (Example; they cited sources like “The Houston Post” and “San Fran Chron,” which are fake news sites.)

• The fake info included false claims about Trump’s Mar-a-Lago wiretap and Ukrainian interference in U.S. elections. (Geez.)

• Most of the fake news sites were (supposedly) created by one man; John Mark Dougan, a former Florida deputy sheriff. He is Moscow now, btw. 🫠

• He had around 167 fake news sites mimicking local news outlets. And AI chatbots couldn’t tell the difference between these and real news sites.

• The tested chatbots included ChatGPT-4, You.com’s Smart Assistant, Grok, Inflection’s Pi, Mistral’s le Chat, Microsoft’s Copilot, Meta AI, Anthropic’s Claude, Google’s Gemini, and Perplexity’s answer engine. It’s only… literally everything.

Why it matters: Turns out AI is easily deceived. Just try asking AI about stock prices or today’s news. It often cites old or outdated articles as if they were current and true. This problem will (probably) be fixed eventually, but for now it shows why we need proper citations and sources for training AI.

This goes beyond news since there’s plenty of misinformation online about even basic things (fitness and health 👀). If you’re using AI to write an article about pool cleaning then make sure the sources are reliable! Don’t believe everything you read.

🔔 The Pope just raised the alarm about AI. Pope Francis addressed the Group of Seven (G7) summit, urging leaders to prioritize humanity in the development and use of artificial intelligence. He was especially against letting AI make critical life and death decisions (so no SkyNet I guess). AI is really stealing the spotlight on the world stage, and everybody’s pushing for regulations.

🤑 DeepMind is transitioning from research into commercial products. Google is combining their AI teams into a single “super unit” (Google DeepMind) focused more on profitability over pure AI research. A few years ago I remember seeing DeepMind’s AI tackle top videogames and the game of Go, beating top players in each. The whole point was to increase the capability of AI. But now Google is focusing their efforts on creating AI products. Bloomberg reported on an interview of two dozen employees who expressed concerns over imposed guidelines and a sense of fatigue, exacerbated by challenges with the Gemini AI model. We’re seeing this same trend happen with OpenAI as well, with ex-employees stating how Sam Altman became obsessed with dangerous advancements and profitability over safety or research.

🍟 McDonalds ends their AI-powered drive-through and partnership with IBM. The biggest reason was botched orders. Yeah, turns out AI was creating more problems than solutions. But don’t champion human workforce yet, this isn’t an AI problem, it’s an application problem. I expect these features to return in a few years once the tech has advanced further.

📱 Gemini’s mobile app launched in India. Google just brought a Gemini dedicated app to India’s mobile market, with support for multiple languages. There’s been a lot of criticism from Indian politicians over fake news coming from Gemini, so I expect this version to be heavily censored.

😵 1 ChatGPT user can replace 50 people

What happened: Freelancers have recently found a new job; making AI sound more human. The problem is this has become the MAIN job available for many freelancers. Some companies have replaced over 50 people with one person using ChatGPT. And it’s a sign of things to come.

This has become especially apparent for writers. AI does nothing better than string together words and paragraphs. So writers have been on the frontlines for the past year. It doesn’t matter if it’s blog articles or copywriting; AI is taking over.

The problem is that AI can produce very standard results. It’s good enough in the eyes of the company. The results are not remarkable… but most people are mediocre at most things, including writing. So most employees would produce standard results anyway (especially on the cheap end). AI just cuts out the middle man.

Freelancers have been the first casualties. The problem is the work involved; most clients need simple tasks done quickly and cheaply. For every freelancer making big bucks by writing for a huge company, there are a thousand others doing menial tasks for $10 a piece. Often these jobs are posted by smaller businesses (or content mills). But such simple work is easily replaced by AI.

The result is a new job category: make the AI output sound more human. A human editor. Low end copywriters have almost entirely moved to this category. The AI creates the sales copy, and the “copywriters” edit that copy. It decreases the writer’s pay and their enjoyment of the craft.

This exact situation is coming for the white collar 9-to-5s. White collar jobs are becoming like blue collar factory jobs. Especially as AI advances and becomes more capable. The machine does all the work while the human steps in now and then to correct the machine. It’s soulless… but efficient. And companies love the cost savings.

Granted, this won’t help small businesses as much as they think. Large businesses use AI for automation or as think tanks. They use AI for things beyond content creation. But small business owners think AI is their saving grace.

Blog owners think, “I’ll use AI to create 100 posts a day. That’ll rack in the cash!”

But everyone can do this. You aren’t rising above the competition; you’re competing on the ground floor. AI (and business success) is about how you use it. If two blogs use AI to create articles on the same niche, what makes one special over the other? What’s their unique proposition?

Same with sales copy. If you and a competing product both use AI… then whose sales copy is better?

Look at the most successful businesses in each arena and how they use AI. These are your guiding lights. Because they are using AI. YOU should be using AI. But they aren’t creating 100 articles a day or sales copy entirely written by AI (even if they’re edited by humans). It gives them no advantage over the competition.

Why it matters: Way I see it you have three options; either learn a valuable skill that isn’t replaceable, do something to become irreplaceable in your workplace (fulfilling multiple roles and designing systems), or become the person using AI who replaces those who don’t.

(We’ll be doing some research into jobs that are more secure against AI.)

💰 AI is changing how ads work

What happened: Google and Bing are using AI more and more in their search results. This is heavily impacting news sites and blogs (mostly for the negative) but it ALSO changes how ads on these platforms work.

The details: 

• Google’s AI overviews often don’t show the ad at all in the answer. In fact, the website they’re citing from also suffers and doesn’t get traffic.

• Bing mostly directs users to informational content now rather than ads.

• Google plans to test ads in AI Overviews. Right now ads suffer more than before in Google search.

• Early tests show ads in Bing Copilot are sometimes irrelevant, such as ads for backpacks appearing in ice bath searches.

• Microsoft in particular is vague about when ads will appear in AI answers, creating problems with measuring ad metrics and response rates.

Why this matters: Users might champion the lack of ads in AI answers… but Google and Microsoft want to make money (look at the figure above). They’ll find a way to incorporate ads, but things will change. These ad networks are also a fantastic way for small businesses to gain traction (we’ve used ads too btw). So the functionality of the ad network is of big importance.

🥊 Perplexity tries to make Google obsolete

What happened: Perplexity now displays the results for factual queries, stuff like the weather, temperature, simple math, and currency conversions. All in an effort to make Google search obsolete.

The details:

• Perplexity now presents direct answers for basic queries with some visual style.

• The info is displayed like a card, similarly to Google’s own quick answers.

• There’s been controversy over Perplexity showing website content without having to access the site itself. An example is Forbes claiming Perplexity shows their paywalled content. (Same with Google too.)

Takeaway: I think this type of info-scraping will create serious problems for websites and publications. There’s talk of paid deals between major publications and AI companies for access (several deals have already happened), but that just squeezes out the possibility for the little guy to shine even more. It’s almost like… the internet is becoming a walled garden.

👺 Venom unleashed!

Source: Midjourney

Give us your feedback!

Got feedback? Rate today’s newsletter by clicking below! 

Got more personal feedback? Fill in the box when you rate us and give us your criticism and feedback.

Thank you for reading! 

❤️Share the (ARTIFICIAL) Love!

Got a friend who wants to appear as smart as you? An AI fanatic or someone shaking in their boots of getting replaced? Share us by clicking the link below and you’ll get our love ;)