Home PublicationsData Innovators 5 Q’s for Lior Hakim, CTO of Hour One

5 Q’s for Lior Hakim, CTO of Hour One

by Eline Chivot
Lior Hakim, CTO of Hour One

Editor’s note: The accompanying image is of one of Hour One’s virtual characters.

The Center for Data Innovation spoke with Lior Hakim, chief technology officer and co-founder of Hour One, an Israeli company that develops technologies for creating lifelike digital replicas of real people. Hakim discussed how Hour One’s realistic synthetic video can improve human-to-machine communication in sectors like education and customer service.

Eline Chivot: What is your background and what has led you to start Hour One? What was the idea behind creating “virtual beings?”

Lior Hakim: You could say that the common thread in my background is machine intelligence: I did my compulsory military service in my country’s intelligence agencies and later spent time both at the Bezalel Academy of Art and Design and at the Hebrew University of Jerusalem where I studied human and machine intelligence in the fields of design, physics, and mathematics. I’m very passionate about the relationship between humans and machines so artificial intelligence was a natural progression for me. Hour One is a platform to explore and play out this passion.

At Hour One, the fascination with virtual beings stems from our obsession with identity. In the case of Hour One, where we turn real people into virtual versions of themselves, we are in fact extending a person’s ability to be in two places at once—in a physical place and a digital one.

We believe that virtual beings can significantly enhance the relationship between humans and machines, by making this relationship more life-like and personable within the digital sphere, which is becoming more prominent in humanity. This is hugely beneficial for brands and businesses who want to communicate with their customers in a more personal and human way, in the digital realm. With our digital human likeness we are making service with a smile a reality beyond the realm of physical human communication, to improve it.

Chivot: Concretely, how does Hour One work and use AI to create synthetic characters? Do you use other technologies?

Hakim: Hour One is a digitization company, we use cameras and microphones as an entry point to the digital realm and then use AI to give life to static inputs. We use advanced neural networks, machine learning, and audio-visual pipelines to create synthetic characters that look and sound like normal people. Once we have a character onboarded, we can take any text input in almost any language and breath life back to it in the form of a living being. We use a vast cloud infrastructure to maintain those beings. This way, a single character can deliver thousands of text lines to a video stream which people can interact with. We then use interaction data as feedback to improve those experiences.

Other technologies have potential for what we do as well. In particular, blockchain is a dear subject to me as I have worked in this field at its early stages. In 2012 I co-authored the Colored Coins white paper which paved the way for multi-currency systems and later became known as ICO (initial coin offering). The meeting point of digitizing quantifiable assets and synthetic characters is closer than one might think, and this is something we are actively working on.

Chivot: Hour One is involved with companies in a surprising diversity of sectors, such as ecommerce, education, automotive, gaming, and communications. Can you give an example of how these companies are using your system for their business activities?

Hakim: Our partners are typically companies that are obsessed with delighting their customers. Let’s take education for example. Education providers that decided that text-based learning is not necessarily the best way to learn are coming to us looking for a solution to transform their entire curriculum—transforming a huge text-based corpus into a video-based platform. It is as simple as the difference between learning from a textbook and in a classroom. Our solution is particularly well-suited for places where there is demand for personable interaction at a very high scale. An educational institution is an entity containing vast amounts of information and usually has a bandwidth issue that limits the transferring of this information to students in an engaging fashion. We partner in a way that enables the end students to get all the education they need with their favorite teachers, without the need for those teachers to recite unfathomable amounts of information, while the teachers behind the characters remain intrinsically involved as well as compensated for providing their likeness.

Chivot: While the technology behind it is extremely beneficial and valuable, there are concerns over deep fakes, and expectations that it will soon become impossible to distinguish between humans and machines. How does Hour One work to address these concerns and to provide a safe use of synthetic characters?

Hakim: We see this technology as an enormous opportunity if used correctly, in an authorized fashion that benefits the human beings behind the synthetic character. We see it not as a race to replace humans with machines but rather to extend human freedom, by enabling them to extend their presence in the digital realm. We absolutely see humans and machines as friends rather than enemies. We plan to make this partnership a long lasting relationship. It may indeed become impossible to distinguish between humans and other—digital—agents and this may raise new questions about our own nature and the nature of intelligent machines. At Hour One, we take this very seriously and are in constant discourse and dialogue about the implications of our technology. Our AI ethics committee meets regularly to discuss this subject and to keep it close to our hearts and minds. We have a number of AI safety measures to protect both our consumers and our characters.

As the leading provider of trusted synthetic media today, we are aware that we need to set an example for others to follow. We are working to educate people about this new technology and its implications. Every asset that leaves our digital doorstep is visually marked as synthetic, and we are promoting the “Altered Visuals” sign which marks the use of our characters. We have reached out to other industry leaders to create those types of standards to keep the audience safe and informed. On the character side, we have done extensive legal work to make sure characters are protected to the extent that is currently possible in such a developing field. We pay our characters for each asset they take part in and they can decide to stop appearing in content at any time. We are also building a system which will enable our characters to affiliate themselves with types of content they relate to.

As pioneers in this field we are most interested in hearing what people have to say, and keeping this important discussion alive as the field matures.

Chivot: How do you expect the technology will evolve to allow you to create characters of even higher quality, ever more personalized and realistic?

Hakim: We expect all these trends to continue at a fast pace, and at some point we will find ourselves interacting with virtual beings through a variety of communication mediums, whether it be through screens, AR/VR, or holograms. At Hour One we are building the architecture.

Today, we’re just scratching the surface. Extreme speeds and high quality will mean that human interaction will be in abundance. Customer service agents will appear instantaneously. We are also excited about the possibilities for job creation. Think of managing your character’s presence, with different personas representing you and working on your behalf, simultaneously, in different contexts. The technology will evolve to become real-time and seamless, and the cost of an interaction will fall dramatically, like sending a text message.

You may also like

Show Buttons
Hide Buttons