Where will AI take us next?
Whatever the next ten years bring in the world of tech, artificial intelligence (AI) will likely be at the heart.
According to Microsoft, 75% of workers who share information as part of their job are now using AI to help them. Consultancy Deloitte’s consumer trend research found that more than 18 million people in the UK alone have used a large language model such as ChatGPT or Gemini.
From healthcare to government, finance to fishing, AI is changing the way society is shaped. Few companies will get far without adding an AI component to what they are offering.
Trying to understand how AI will change in the next decade could be considered a fool’s errand, given the leaps the tech has taken since ChatGPT burst onto the scene in 2022. Large language models (LLMs) are changing the way we interact with content. AI in healthcare is speeding up drug discovery. AI vision enables self-driving cars, interprets geospatial data from satellites, and allows farmers to predict what crops will fail. And these are just some of the many use cases that exist today, while hundreds more are yet to emerge.
With the advent of AI being driven by businesses and consumers alike, to understand what the next ten years could look like, we sat down with Daniel Verten, Head of Strategy at Synthesia. One of the UK’s leading AI start-ups with a valuation of over $2.1bn, Synthesia turns text into AI-generated videos in over 140 languages, and has been adopted by major global organisations including Reuters, Zoom, and Heineken. In addition to his role at Synthesia, Verten serves as a member of the World Economic Forum’s AI Governance Alliance, which is committed to fostering inclusive, ethical and sustainable AI, giving him an important perspective on the technology’s potential and the hurdles that must be overcome.
For Verten, AI video has almost limitless potential.
“The interesting thing about turning video production into a software-only process is that it allows you to create new formats and experiences that previously were not possible. AI is driving a fundamental shift in how we produce and consume video,” he told TFD.
Clients ranging from big corporations to smaller businesses use Synthesia to create video content and avatars for use in a range of scenarios: training videos, chatbots, product demos or ads. For example, PepsiCo created an avatar of footballer Lionel Messi and used it to send personalised messages to fans in up to 10 languages.
Creating Hollywood-quality content on your laptop
When it comes to how AI video will develop in the next ten years, Verten acknowledges the speed of change.
“I’ll be very honest, I have no idea because we haven’t invented it yet.”
He does, however, have one bold prediction.
“Just as the internet went from being static web pages to a full multimedia experience, so too will AI video. I’d say that on a five-year horizon, anyone with a laptop and their imagination will be able to create Hollywood-grade content on their laptop from their work desk.”
And if that sounds disruptive, then he is drawing reassurance from the past.
“It’s useful for us to study history and look at previous platform shifts. The first digital cameras were invented in 1975, but it took 14 years for the first commercially available digital camera to hit the market. While most people think that new technologies replace old ones, that’s not actually the case. What really happened is the per-minute cost of analogue film fell to about $5, and that allowed the market to expand exponentially because all of a sudden, more people were able to afford a camera.
“The same thing is happening with AI video – instead of a few select people with access to expensive filming equipment being the only ones able to create high-quality content, I believe AI will democratise content creation, not replace it.”
Post-truth era
This is the optimist’s view, but many more are pessimistic about where the new age of AI-based content creation will take society. Already, deepfakes have been used to scam people, create pornography (including images of child abuse), and disrupt democracy.
One new trend is the stream of so-called “AI slop” that is filling social media channels. Some are clearly fake videos, such as whales flying through space. Others are more ambiguous, like a polar bear being hauled onto a fishing boat. Comments below such videos suggest that, as a society, we are already facing a new digital divide – between those who believe what they see and those who are starting to view all content with a critical eye.
“Education will be key,” said Verten, as we enter what some have called the post-truth era.
“At the start of the conflict in Ukraine, someone created a deepfake of President Zelensky and hacked broadcast TV stations and distributed the video on those channels. But it was actually debunked by the public in five minutes, and President Zelensky shot a video of himself and put it out, saying the other one was not me. With proper digital literacy and public education, many of the risks can be mitigated.”
He does think it is important that people play around with the tools.
“I think as long as it is harmless, I actually see it as a good thing. Because one of the novel things about generative AI is that you can get started right away and experiment with it. And that helps educate the public on some of these new technologies in a playful way,” he said.
The interesting thing about turning video production into a software-only process is that it allows you to create new formats and experiences that previously were not possible. AI is driving a fundamental shift in how we produce and consume video.
Consent, control and collaboration
Deepfakes for more nefarious purposes will need tough regulation, and already we are seeing nations outlawing the making of deepfake pornography or images of child abuse, he said.
Ethics has always been a watchword in the world of AI, although finding international agreement on what that means, and making sure the tech giants have appropriate guardrails, is less easy to enforce.
There have been disturbing reports of AI chatbots calmly telling a man the best method by which he could kill himself, just one of many shocking examples of how AI may be running wild.
Verten thinks that AI systems, whether developed by big tech companies or start-ups, need ethical principles at their heart, and in the case of Synthesia, this is a simple mantra: consent, control, and collaboration. Consent means no AI avatar is created without the consent of the person whose likeness is being used. Control means that no one can create a video that is against the firm’s terms of service, and collaboration sees the firm partnering with governments and organisations such as the World Economic Forum to “help build a healthier media ecosystem”.
In a world where our images could become like a form of currency, Verten thinks that change is afoot.
“I do believe that we will have to bring regulation into this new age and give people protection and agency over their likeness.”
Some of the technical solutions being mooted are more sensible than others, and he is not a fan of the move to digitally fingerprint all AI-generated content.
“I think that is a ludicrous exercise. What we should do is exactly the opposite and start to tag trusted sources, such as the BBC, with a cryptographic signature.
BBC Verify uses this already, so I can be 100% sure that the content came from the BBC and was not generated by someone else.”
In future, Verten thinks that all politicians and business leaders will get their own cryptographic signature to validate content they are sharing.
Instead of a few select people with access to expensive filming equipment being the only ones able to create highquality content, I believe AI will democratise content creation, not replace it.
Will AI outperform humans?
One of the biggest worries around AI, as it increasingly makes work more efficient, is how quickly it will render human workers redundant. Again, the world is firmly divided between those who see the age of AI leading to mass job losses and those who see it working in harmony with humans for many decades to come.
A report from The Institute for Public Policy Research predicted that eight million UK jobs alone could be lost to artificial intelligence in a “jobs apocalypse”, while Bill Gates said that advances in AI mean that we will no longer need humans “for most things”.
Verten takes a more optimistic view.
“60% of the jobs that we do today did not exist in 1940. It’s always easier for us to think about the jobs that might get disrupted. It might be a bit of a cliché, but you’re not going to be losing your job to AI, you are going to be disrupted by another human who knows how to use these AI tools.”
One big question being asked currently is how advanced those AI tools will get, and there are mutterings that the leaps Generative AI has made in the last three years may soon hit a wall.
Verten doesn’t agree.
“Once Apple sold iPhones to a billion people, the improvements became incremental, but we are very, very far from that in AI. So I think that we are very far from the point where it starts to plateau.”
Creative potential
The billion-dollar question in the AI world is when and if we will get to a point of Artificial General Intelligence (AGI) where an AI is as intelligent as a human, and able to multitask in the way people do. This has been the holy grail for AI for many decades. Google DeepMind CEO, Demis Hassabis, said such a system could exist by 2030.
Verten prefers not to speculate on this but is characteristically optimistic about the future of AI-generated video.
“What I’m super excited by, and one of the reasons that I get out of bed in the morning, is the new formats and creativity that AI video will unleash. Similarly to when we got smartphones, people started to experiment with all sorts of weird and fun ways to express themselves. Seeing that creative energy and being able to enable creativity for so many more people, I find it magical and really, really exciting in the next few years.”
Curious about TFD's work in AI tech? Learn more about our AI Collective: https://www.wearetfd.com/collectives/ai-collective
