Imagine scrolling through your feed on Tiktok or Instagram. You get stuck on a particular clip. It's a trailer for a movie that looks interesting, even reminiscent of something you like. But the movie doesn't exist. The trailer is AI-generated. Created using Google's new Veo 3 tool, where you can easily write a prompt in a few seconds and get a realistic clip of what's being requested.
Ever since Chat GPT was released, humans' play with reality has accelerated, says Andreas Önnerfors, media researcher at Linnaeus University and project leader for the media institute Fojos Faktajouren.
But now it's so realistically simulated that you can talk about a second reality that mirrors our physical reality.
Virtual Reality
Other tech giants besides Google are constantly working on sharpening their AI tools. Sometimes there's a leap in realism or accessibility, like with Veo 3, and within hours, special forums on the internet are flooded with people testing the new technology's limits.
What Veo 3 creates doesn't have to be a fictional trailer for a movie that never will be. It could just as well be a beautiful goal in a soccer match that never took place or a shocking quote from a political meeting that never occurred.
We'll see political candidates who can speak in 150 different languages. And it's possible that entirely virtual people enter our lives. We're already seeing digital models and news anchors in some places, says Andreas Önnerfors.
At the same time as the development is generally moving rapidly forward within AI, there are question marks around how much better tools like Google's Veo 3 can actually become in a short time.
There's a discussion among those who research AI about whether AI has hit a wall. That is, how much more do we need to pay to get a service that's marginally better than what we have now, says Ariel Ekgren, researcher at AI Sweden.
Historical Comparisons
Ekgren agrees, however, that what's coming out now with Veo 3 represents a new standard for speed.
Just a few years ago, these clips were incredibly unrealistic, now they're starting to approach something that seems realistic. The fact that they're now doing audio as well is impressive, he says and continues:
There are historical comparisons to be drawn when it comes to the consequences of technological development. If we look back at, for example, landscape art, then came film, and later moving images. Each step brought new opportunities and challenges.
Major Political Issues
But aren't all AI tools and services just tools and services, limited by how people use them? Yes and no, according to Ekgren and Önnerfors.
The technology will be able to do your bidding, both for campaign purposes and for manipulative purposes, Önnerfors and adds:
We risk ending up in a mirror world. Where the digital and virtual become increasingly important. But just as likely, there could be a backlash, a disconnected movement.
The use of these services will quickly become major political issues, says Ekgren.
With the development pace, the debate about how AI affects society as a whole will always come up. If it turns out that AI leads to better education and innovation, and we want to use AI more, we'll also need to invest more in information and energy infrastructure. Is there sufficient understanding of AI and investment willingness to meet the future?
Artificial intelligence (AI) is intended to mimic the human brain's ability to acquire knowledge, make conclusions, plan, solve problems, and interpret impressions.
AI has become increasingly common in various parts of society and everyday life. AI is also commonly found in fiction and often takes a form that is far from where the technology is in reality. A human-thinking and acting robot, general artificial intelligence, is considered scientifically impossible.
Chat GPT is the most successful AI service. The service answers users' questions and can help with everything from dinner tips to generating images.
Veo 3 is Google's latest video generation service. Users write in what they want to create (prompt), and shortly after, they get the result in the form of a few-second-long video clip with audio.
Source: National Encyclopedia, Nature