03/22/2023

Former president of the United States Donald Trump was not arrested on the streets of New York nor is he in jail As seen from dozens of photos shared on social networks in the last few hours.

Although the pictures closely resemble real-life scenes, they are all fake. It is a photo created with the help of Midjourney v5 artificial intelligence (AI) which produces such precise cutouts that it reaches an extraordinary level of detail: it is very difficult to distinguish whether it is true or false. This is a system where anything that is requested from text can be converted to images (text to image).

Going back to the Trump case, there are photos, for example where The former president is seen wrestling with several police officersand elsewhere he is wearing an orange suit behind bars in prison.

The person responsible for sharing the content that has generated all these kinds of opinions is British journalist Eliot Higgins, who does it on his Twitter account.

The facts add several problems to the AI ​​debate: where are its limits, how far can it go, how can it influence the perception of reality and create a “reality” that doesn’t exist.

Because Artificial Intelligence develops, learns and allows for more possibilities. According to the creators of Midjourney, the latest version of the tool has, among other improvements, “significantly better image quality, a wider variety of outputs, a wider range of stylistic cues, support for textures, a wider range of image formats, and a wider dynamic range.” That is, the more realistic photos that can be produced.

it is over

In 2017, he did too Former US President Barack Obama was a victim of artificial intelligence. Researchers from the University of Washington created, applying AI, a digital video in which a politician speaks with his voice about some of his presidential accomplishments. He simulated his mouth with everything he said. Experts combined archival images of former presidents with audio that mimicked Obama’s voice, like an impersonator.

To produce this content, a technology is known as multimodal generative artificial intelligence: combine text, video, photos, and even programming languages. This is very different from those used to make predictions and classifications.

Antioquia University Professor, John Fredy Duitama, explained that this type of technology is nothing new. “It’s been around for a while, all we’re seeing is that it’s on the front page,” and it’s getting more and more accessible to the everyday person. In the list is Dall-e and Stability.ai to generate images from wordsBooth.ai to take quality photos in seconds or GitHub Copilot to help you program.

Anyone can log on to Dall-e, for example, to have him create a picture they like: a cat on the moon eating ice cream pop-style. And although there is a paid version, it is possible that it is free. In the example there are free credits.

From the audience’s perception, for Karina Ochoa, a teacher at La Colegiatura, This type of content has a very strong impact on society.especially when disseminated on digital channels and platforms such as social networks.

“AI as a tool it can be the best assistant to eliminate repetitive processesSo we need to start talking about the ethics and morality of AI to use it responsibly.”

There are two keywords: ethics and morality.

Supervise!

How do you know if an image is real or not? This is one of the big questions out there today. Guru Duitama said that it would not be easy for unsuspecting people to find him, because behind everything there is an underlying problem and it is a position that receives information. Something similar to what happens with the fake news that haunts the internet.

But calm down. His vision was not as dark as it seemed. The teacher explained that this time there are tools that can help detect these “deep fakes”. or content manipulated by AI. One is FakeCatcher which offers 96% efficiency.

“At first glance it is very difficult to recognize this type of image because it is very well simulated and to this is added the position of the person looking at it, whether they think it is correct or not,” added the professor.

He also stated that all technology can be used for good or bad. In education for example, AI can be very useful for creating personalized content for students as a version of the story according to the interests of the reader.

The downside is the AIe can be used in many negative ways for fraud (code to attack servers) and promote fake news, images (as in Trump’s case) and videos with the aim of “burning the nation”.

“It is very important as a consumer to be aware of and have a perception of what is consumed on social networks, be it brands, experiences or affiliations with entities or politicians. It’s important to always reinforce information, don’t be driven by likes or because the topic is controversial,” said Ochoa.

On regulatory issues, there has been some progress. However, the ChatGPT robot, created by the American company OpenAI, has put the European Union (EU) at a crossroads in its efforts to regulate the use of this technology to protect users.

For more than a year ago, The EU has drawn up regulations that will dictate the use of AI: advances in the framework for regulating this type of technology in an effective manner. In April 2021, the European Commission first announced a blueprint for a set of AI rules, and the European Parliament hopes to finalize a draft of the AI ​​law this month. In the case of Colombia, regulation of artificial intelligence is still nil.

new announcement

this tuesday, Microsoft also announced the Bing Image Creator tool, an AI that turns text into images. And last week, the tech giant made another big announcement: It’s adding its generative AI to some of its most popular business apps like Word, PowerPoint, and Excel.

Google doesn’t want to be left behind: It says it opens up early access to select groups of users in the US and UK to interact with Bard, its AI-powered chatbot and rivals Microsoft’s ChatGPT OpenAI and Bing.