CHATGPT PLUGINS

🖥️ Is this the new way we’ll use the internet?

Users who pay for ChatGPT Plus will have access to a web-browsing feature that will provide up-to-date information.

This is a huge step in making ChatGPT a search engine. Prior to this, it was only trained on data that dated back to 2021.

With ChatGPT's web browser, now, selected users can ask multiple specific questions and the bot will search the web, summarize the answer, as well as multiple news articles it deems relevant in a matter of seconds.

My opinion: It seems like Bing and Google are getting a competitor. However, I am sceptical whether we want to browse the internet through a chat window for less complicated queries. ChatGPT has to get a better UI.

About the plug-ins. OpenAI is also rolling out 70 third-party plug-ins for web services like Expedia, Kayak, and Instacart.

From now on you just type in “give me a good quinoa salad recipe and order the ingredients” and pay for it in two clicks.

With the plug-ins, users will be able to find cheap flights, practice languages, plan summer vacations and discover new restaurants.

As soon as I get access to these features I’ll report back hopefully from an exotic location I’ve found a ridiculously cheap ticket to.🌴

BAD COP NO DONUT

👮 AI tools used by police who 'do not understand how it works’

A troubled police officer, leans closer to a screen in a half lit room, holding back his breath as he hits enter. The computer works its magic and enhances a pixelated license plate.

Gotcha! - He says as he hurriedly jots down an address from a database and runs away grabbing his pistol holster on the way out.

This is how every TV show imagined “state of the art” police technology a decade ago, however, AI is rapidly surpassing these once miraculous capabilities, in fact it’s already getting used in real-world investigation for facial recognition and gunshot detection. Which is all good but…

Quote: “The study participants [cops] were not familiar with AI, or with the limitations of AI technologies.”

Why is it a problem? AI algorithms are not foolproof and according to a 2019 study they are prone to be biased. They falsely identify African American and Asian faces 10 to 100 times more than White faces. If officers are not aware, they may trust it more than they should.

No disclosure. On top of that a recent article by the New York Times shed light on a case of wrongful arrest due to facial recognition. Notably, the court documents and police reports from the case lacked any mention of the use of AI technology - a trend that is reportedly on the rise.

Cops “should have at a minimum a broad understanding of the AI technologies.” - the NC study concluded.

CHATGPT VS TRUTHGPT

✏️ Writing with AI help can shift your opinions

Historic background: During the Korean War, a shockingly high number of captured American soldiers collaborated with the Chinese Communists.

How did the Chinese achieve this?

They asked the prisoners to make statements that were so mildly anti-American or pro-Communist that they seemed inconsequential. For instance:

“The United States is not perfect.”

Subsequently, they would urge them to come up with some examples and finally write down these “problems with America” and sign it.

These lists were then broadcast across the POW camps through radio, exploiting people’s inherent need to stand by what they said. As a consequence, these prisoners' opinions and self-images changed to become more consistent with their written words.

Something similar is happening with AI. A Cornell University study found that writing tools with inherent biases might shift our opinions. Researchers steered a large language model to have either positive or negative opinions of social media, then they asked participants to write essays with it.

The result: People who co-wrote with the pro-social media AI assistant composed more sentences arguing that social media is good, and vice versa, compared to participants without a writing assistant.

The survey revealed that a majority of the participants did not even notice the AI was biased and didn’t realize they were being influenced.

The bigger picture: Just as social media has helped the spread of misinformation and the formation of echo chambers, biased AI writing tools could produce similar shifts in opinion, depending on which tools users choose. For example, Elon Musk recently has announced he plans to develop TruthGPT, an alternative to ChatGPT, designed to express more conservative viewpoints.

Tools worth Trying 🔩

• Magic Editor: a new AI toolkit that performs complex photo edits inside Google Photos.

• MusicLM is Google’s text-to-music tool that turns text descriptions into tunes.

Ai Art of the day 🎨

Check out this Coca Cola ad that was partly created by Stable Diffusion, a popular open-source generative AI. Although it was just one of the many tools used to create the commercial, it's still a significant step towards mainstream adoption of AI-generated visuals.

🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄🌊🏄

That’s it folks!

If you liked it please share it with a friend and make this writer happy that ChatGPT didn’t replace him yet!

Or did it?