Visual AI is indefatigable, tirelessly working endlessly on tasks like tracking medical procedures, and infinitely scalable, because once we replicate a visual task like counting and typing cells under microscopes, we can simply add more GPUs and count millions of cells per minute. Visual AI can do anything the eye can do, and better, says Jeffrey Goldsmith, Vice President, Marketing at Chooch
How is visual AI a neural network of perception? And is it accurate enough?
Those are two separate questions – what is visual AI and is it accurate enough. Visual AI isn’t a neural network, it’s an entire process, from visual data, to training, to generation of models and neural networks. Then, of course, there’s deployment. Is the visual AI to be deployed purely in the cloud and connected to incoming visual data via API? Or, will we deploy our inference engine to the edge, on an NVIDIA Jetson Nano, for example. Visual AI is a complete solution, allowing visitors to a zoo to use Visual AI in AR to identify animals and plants and access information about them. As far as accuracy is concerned, we have an extremely fast way of importing 1000s of images per minute. One of our customers uses us to check images from their dealer network before the images are posted on their retail platform and our Visual AI platform recognizes anomalies with very high accuracy.
How is facial recognition combating the cybersecurity threats?
We actually don’t focus on facial recognition, we focus on facial authentication. We answer questions like, are you supposed to be boarding this plane or are you the patient who is schedule to get an appendectomy. We can lower the risk of medical procedures being performed on the wrong patients and we can prevent people from accessing spaces and data and financial information by authenticating them with 512 biometric features and liveness detection. That way, you can’t use a picture or video of Jeffrey Goldsmith’s face to get access my top secret information.
Recently, the EU digital chief claimed that automated facial recognition is a GDPR breach. He stated that facial recognition in the EU should require consent. What are your views on this?
I agree with that. We all want security, and I think there’s an incentive for people to say, “Yes, I agree to use my face for authentication. My face needs to be in front of the ATM camera for the bank to dispense funds.” I don’t think there is any incentive for anyone to say, “Yes, I want the government to know where I am at all times.” That said, we will soon have 1 billion cameras at street level all around the world. Obviously, those cameras will be enabled for AI on the edge. The entirety of humanity would have to watch streams of all of those cameras if we wanted to get any value there.
Which industry you feel has witnessed the maximum adoption of biometric and facial recognition technologies globally?
SSO, which makes total sense because we are all signing on to multiple services every day. It’s a staggering need, and this could eliminate bots if Twitter and Facebook adopted a one person-one account policy and required true personal authentication. I can’t believe Docusign isn’t doing facial authentication. I could sign your documents and you could sign mine. Financial information is obviously critical because there’s an economic incentive to protect your money. Then there’s physical security, and one of our customers uses us to secure remote locations. That’s a very informal survey. Faces are completely unique and as long as your solution has true liveness detection as ours does, then it’s super secure.
Jeffrey Goldsmith is Vice President of Marketing at Chooch, the leader in Visual AI for the enterprise. He’s been head of marketing and consulted for many startups. Jeffrey has also written for a wide variety of publications, from Wired to Conde Nast Traveler, and during his advertising career, he created global advertising for Wells Fargo Bank, Sun Microsystems, Hilton, and many other brands.