A.I. is not all bad — like the internet it will cost jobs as well as create them, but we must watch it like a hawk
WILL AI take my job? This is becoming a common question on Twitter, Google and in the local pub.
Anxiety around AI technologies has been advancing at nearly the same pace as the technologies themselves.
In a recent survey we conducted with pollsters Find Out Now, 68 per cent of Brits had begun feeling worried about AI. While 63 per cent felt it could have devastating consequences for humanity.
It is clear that concern over AI, and our future lives with it, are at a high. So let’s try to understand what we know now.
Artificial Intelligence is already starting to take a noticeable toll on jobs around the UK and globally.
IBM paused hiring for roles it believes could be replaced by AI tech in the coming years. Most impactful yet, BT has outlined plans to cut around 55,000 jobs by 2030, 10,000 of which could be replaced by AI.
READ MORE ON ARTIFICIAL INTELLIGENCE
Positive impact
As someone who was doing a PhD on blockchains during the crypto-hype cycle in 2018, I see many parallels between crypto then and AI now.
There is no doubt AI will change many aspects of our lives, but we need to be cautious against both utopia and doomsday scenarios.
AI is not all evil, it will likely create many jobs, just as other technologies have before it, and it has already been used in ways that have immense positive impact.
MIT researchers created an AI model, Sybil, that can detect future lung cancer risks and help with early diagnostics. Technologies like this have the potential to save hundreds of thousands of lives.
Most read in The Sun
Deepfakes can be used to modify the mouths of actors in movies, making dubbed movies much more immersive for non-native speakers. However, like any great technology, such as nuclear, AI can be a weapon of mass destruction.
Deepfakes are far more commonly used for malicious reasons, pretending to be somebody they aren’t. More than 95 per cent of deepfakes are accreted to produce non-consensual pornography.
An AI-generated picture of a woman was shared on Reddit, scamming viewers into paying her for illicit content. Claudia, as her makers called her, never existed, and all of her photos were generated by an AI model.
We all laughed at the fake AI pictures of the Pope dressed in a white puffer coat and .
But these images are now so realistic they can fool most people. In the wrong hands that technology could have far-reaching consequences.
Russia has been putting out false but highly convincing deepfakes of Ukraine’s President Zelensky telling his troops to lay down their arms. There are reports of banks being deceived by AI-generated voices of executives asking to transfer funds.
Online spaces that require belief to function risk losing that when we can no longer trust what we see with our own eyes.
Two-thirds of Brits polled believed we needed to authenticate all image and video content on social media. We agree, which is why we started building frankliapp.com, a platform that ensures every photo and video on it is genuine.
Safeguarding online spaces as a place to consume trustworthy information is absolutely vital. However, it isn’t just about building counter-solutions to the threat AI poses.
The biggest risk imposed by AI is created by the environment in which it is developed. There are next to no regulatory guardrails guiding the development of AI and, crucially, the procurement of data to build these tools.
Prime Minister Rishi Sunak has made similar exclamations, discussing the need for guardrails, but so far little has been done in the form of policy action.
Nearly 80 per cent of those we spoke to think the AI industry needs to be more closely regulated.
But for now the future of AI rests firmly in the control of the tech firms that are designing these technologies. They are driven by market competition rather than by what’s best or right for the British people — or the world.
So we live in the AI Wild West.
We need governments to start enforcing privacy regulations to protect people’s intellectual property from being unilaterally captured by these AI models.
Global consideration
These efforts will require global co-ordination.
Our Government should focus on minimising the impact of AI on the working class, as we are already seeing companies who are very keen to cut jobs and increase their profits.
This is bad not just for those losing their jobs, but also for the consumer, as they are being made guinea pigs for an untested technology.
The PM should give greater powers to privacy and data regulators, as AI will touch every corner of the UK — and every aspect of government.
AI will transform our lives, help us live longer and save time and money.
It will cost jobs (some unnecessarily) but create others too. And it will also supercharge disinformation warfare at a scale never before seen.
We need to handle AI with care and keep a hawk eye on it at all times. This threat is not artificial.
READ MORE SUN STORIES
It is here now — and very real.
- Dr Mansoor Ahmed-Rengers is the founder of internet tech firm OpenOrigins.