Me And My Digital Twin: I Have A Brand New Deepfake Avatar

Me And My Digital Twin: I Have A Brand New Deepfake Avatar

I've previously written about the impending era of deep fake medical videos and how to spot if you’re talking to a deep fake physician (or a digital avatar) instead of a flash-and-bone one. As we discussed back then, synthetic healthcare workers will arrive eventually, as the number of missing doctors, nurses and other medical personnel will only increase globally in the coming decades.?

It was also widely discussed on our platforms that with the rapid development of various AI algorithms, we can soon find ourselves in a situation where dealing with a medical AI chatbot will provide less risk than waiting for an appointment with a human doctor.?

Just to refreshen the memories, when trying to spot a deep fake human, look for unnatural eye movements, unnatural facial expressions, hair, teeth or other small details that don’t look real and blurring or misalignment – especially around the edges of the person. Practising our deepfake-catching tricks might become a handy skill in the mid-term future.

But back to our topic: recently, I was approached by Synthesia with an intriguing proposal - would I be interested in creating a digital avatar? Of course, I was! The prospect was too tempting to resist.

Initially, the plan involved me travelling to London to work with their team. However, with my own studio, crew, and equipment, including a 4K camera and a green screen, we were able to tackle the task right at The Medical Futurist HQ. The team and I compiled the necessary footage for the process without setting foot outside our studio.

Synthesia provided me with a precise script, which I had to read out and I also received specific instructions on how to act. That was all the necessary preparation.

Only a few days later I was granted access to a platform where I could start using my digital persona, dictate any text and have it read out by my digital avatar. The platform also allows me to incorporate any type of media, cutscene, or music into my video. It also gave me the option to choose from about 100 different languages/accents.

How do I look??

At first glance, my digital avatar bears a strong resemblance to me, which is a testament to the technology's capabilities. However, after looking at it for a few seconds, you start to feel uncomfortable. Digital Berci with a Scottish accent* is creepy - my team declared unanimously.?

(*Pardon my weird sense of humour - it was my own choice.)

First of all, my avatar doesn't blink. At all. That is by far the eeriest part of the experience. Also, its gestures and emphasis points are slightly off, making it apparent that the technology isn't quite perfect yet. Nevertheless, it's exciting to see what a digital avatar looks like and how it functions.

To demonstrate this, we've been working on a new video that will let you decide how convincing my new avatar is - whether it's me speaking, or my digital counterpart. After a few days, my virtual twin delivered this slightly corporate marketing message, while it has learned how to blink.?

I already see good use cases, even in medicine

The interesting question for us is whether this technology can be applied to medical purposes. For now, the answer is no in the case of replacing live doctor-patient interactions. The "uncanny valley" effect is too strong; the avatar lacks the ability to convey emotions and empathise, which are vital in the medical field. Therefore, at this stage, a patient would likely find a written message from their doctor without a visible avatar more comforting.

However, there are already potential use cases. For instance, I can envision a doctor's introduction video being used to create a series of educational materials, or hundreds of videos being recorded for professional development courses instead of someone spending hundreds of hours in front of cameras. It's faster and cheaper, and if the target audience also consists of [healthcare] professionals seeking new knowledge, this medium can prove to be very efficient.

All in all, this technology is quite new and there’s still room for improvement. Digital avatars need to advance further to overcome their current limitations, but they also hold vast potential. We will need to find their just places and work out the boundaries of how to use/not use virtual humans. Until then, my digital avatar and I will continue our fascinating journey into the digital realm.

Sally Ann Frank

Worldwide Lead @ Microsoft for Startups | Driving Digital Health Innovation | Speaker | Author

1 年

Bertalan - like any technology, these avatars will continue to be refined. What happens then when the line between virtual and real beings is not so evident?

Cheima WAFI

Biomedical Engineer | Master's Student in Microengineering | Driving Healthcare Innovation & Patient Outcomes Through Advanced Medical Technology Solutions????

1 年

thanks for sharing with us , it's sounds that you had a great experience, for me, the combination use of digital avatars and virtual reality is efficient to improve health professional's education by creating simulations that allow nursing students to explore role-play and use computer-generated interactions with patient avatars to enhance learning in a risk-free environment.

回复

"... when trying to spot a deep fake human, look for unnatural eye?... " I guess, this is best done by a mobile phone app, with reporting of fakes integrated into the process of analysis ?

回复

Thanks for sharing this. Very interesting to see how far this has come so quickly.

回复

要查看或添加评论,请登录

Bertalan Meskó, MD, PhD的更多文章

社区洞察

其他会员也浏览了