As we age, it's not uncommon for people to slow down, cease regular exercise, and lose elements of their fitness. Whilst there may not initially be a noticeable decline in health, once an individual gets unfit or unwell, it can be challenging to get their health and fitness back up. Maintaining an active lifestyle is an easier feat than working your way back up to one, so it seems sensible to encourage a certain level of activity as we age. ElliQ is the proactive AI driven social robot trying to overcome this problem by encouraging an active and engaged lifestyle by suggesting activities and making it simple to connect with loved ones.
At the Deep Learning for Robotics Summit this June 28 - 29 in San Francisco, Shay Zweig, the Head of AI at Intuition Robotics who have created ElliQ will be sharing his work on creating state of the art algorithms in decision making, robotic vision, dialogue management and more. Shay explained that the idea of working on a product like ElliQ was very exciting to him, and also a major challenge, given that it is such a complex product that is trying to do something that has really never been done before – learn about its owner in order to personally and proactively suggest ways for him or her to stay active as they age. Since joining the team, Shay has been focusing on two main areas within AI. 'For one, we’ve been investing time in computer vision abilities, trying to make ElliQ understand its environment more effectively in order to make better decisions. The second area of focus is the decision-making engine itself; many of the algorithmic rules around decision making have been defined by experts, so we are looking at how we can combine that information with the data we’ve collected from our own user testing for our unique audience in order to develop enhanced decision making and manage the full interaction. What we’ve ended up doing is developing our own algorithm which is a multi-goal agent: it is trying to accomplish multiple things, and learns from all of it in order to balance and prioritize actions for each user. We need to make ElliQ proactive in a truly smart sense – to know what to do, when to do it, and the best interaction approach, and this is going to be different for every person. It has been quite the challenge, but an exciting one at that.'Social robotics is still a very young industry with limited opportunities, so this is Shay's first foray in the area. With a PhD in Neuroscience, when he began speaking with the team at Intuition Robotics, Shay recognised that this would be the culmination of his previous experiences; 'understanding the brain and how people operate on a cognitive level; combining this with my experience in machine learning and computer vision means I am able to think holistically about the human experience with the technology we are developing.'
For one, the challenge is immense. We are trying to develop these huge platforms and have really only scratched the surface of solving the problems this tech provides. But on the other side of this is the potential impact we can have. The field is so young, and seeing where we are now, I can’t help but think where we can be even fifty years from now. Getting to work on projects at these initial stages is really inspiring to me.
With Intuition Robotics specifically, we are really building an entire platform meant to help people. Aging alone is a massive problem, a scary one, and one that many people are faced with. We see social robots as a means of supporting with communication; these robots are able to mitigate communication with people who may have a problem doing so on their own, perhaps because of social isolation, or a disability. Products like ElliQ could potentially solve a major issue for these people by creating a platform for them to communicate with the world around them, and the impact would be enormous for their lives.
On a bigger scale, we are looking at the inevitable fact that robots will be all around us at some point in the future. If we can define the relationship humans have with robots, and do it in a way that feels authentic and empowering for the human, then we can perhaps create a more idealistic future.
Our biggest challenge is in creating a truly personalized proactive experience. To achieve that we must be able to balance the goals for each user based on his needs. This should all be done using a very limited amount of data and with low computation. The advantage that we have is that our robot has the ability to interact; it can get some information from peripheral sensors, but it can also ask questions. So ultimately, we are building an active learning framework to maximize the information we get while simultaneously maximizing and constantly improving the interactions people have with ElliQ.
I believe that in the not-so-distant future, everything electronic will have an AI component. All of the tools we use, our entertainment, our daily tasks, will be automated. It is our role, in this industry, to make these products something we actually want to live with – to make sure they are actually improving our lives. There is an ongoing ethical conversation to be had around what these interactions look like, how can actually guide our usage of robots and AI in a way that prevents the dystopian future we see in the movies…but ultimately, this is the path we are going down, and I think we have an opportunity to greatly improve lives and help people who need it.
Aside from social robots, we are looking at AI as a means to create proactive user experiences – ones in which technology learns from us in order to proactively make our lives easier, rather than waiting for our commands. This technology could affect any industry utilizing AI, from transportation to entertainment to our homes and offices; if our technology could understand our goals and needs, and then provide the tools for us to be empowered, it could really change the world and the way we live.
I would absolutely recommend a career in AI, it is a diverse and fascinating field with endless challenges. However, I would warn that this is not an easy path to choose. A person who goes into this field needs to not only be a great mathematician and have good algorithmic skills, but also a good understanding of how your product is being used – how the algorithm will fit into its environment. I also think you need to understand software, and how to put our AI into production systems. The best way to apply AI is to be able to think holistically, be able to talk to people on different teams in order to know the context and solve the right problems. It is an immense challenge, but also a rewarding one!