Andra Keay is the Managing Director of Silicon Valley Robotics – a not-for-profit coalition of robotics companies and startups supporting the innovation and commercialisation of robotics technologies. She advises a number of startups and is the founder of Robot Launch Pad – organising events bridging startup and robotics communities.
We caught up with Andra to chat about the future of robotics and how entrepreneurs in the field can ensure innovation remains beneficial to human beings.
How did you get into robotics?
“I’ve been interested in all the geeky things since I was a child. Both my parents encouraged that a lot; their roles at universities meant that I got to play with computers and rockets and build equipment at a really early age.
“But one of the things that happened all the time, was that people would express their astonishment that a girl was interested in things like rockets and robots. And after a while, that astonishment became more interesting than the things that I was doing.
“I studied theories of communication and what I would like to call human-robot culture, as opposed to human-robot interaction. Where human-robot interaction focuses on the psychological side of things, and it’s certainly a fascinating field, I realised that the one-to-one relationships are nowhere near as indicative of what’s happening in society as understanding the broader social shifts.
“It’s not really a discipline as it were, but we have to ask ourselves (and this is the question that fascinates me): what is the 21st century going to look like as the most cutting edge technologies trickle their way through to mainstream adoption?
“We’ve seen what the adoption of technology looked like in the 20th century and that’s not simply the internet or even computing; it’s technologies like the automobile and household appliances.
“If we look to history, we can see a lot of great examples of what happens when a really disruptive technology is adopted. It’s not an immediate thing – as Roy Amara said, we often vastly overestimate the impact of technology in the short term but vastly underestimate the impact in the long term.
“We fill the headlines with fear about robots and AI in the short term, but it’s probably longer term structural shifts in society that are really where both the potential for problems and the potential for great advances are.
“One thing that’s clear, is that to solve a lot of the world’s serious problems (the ageing population, the increasing population and the fact that we need to double food production without having the ability to increase the amount of land that we use), we are probably dependent on technology because we haven’t been able to fix those problems yet.
“We’ve really hit a saturation point. The ability to increase our productivity without using more manual labour; we’ve hit a saturation point there. We’re dependent on technologies now to increase productivity. We need to be in the upward spiral or we’ll be in a downward spiral; there isn’t really a static place where we can just exist comfortably.
“One of the grandfathers of AI, Norbert Wiener, said that the future is not a comfortable hammock for us to lie back in and be waited upon by robot slaves; it’s actually an ever more challenging fight against the limitations of human intelligence. That’s really at the heart of my interest in robotics and AI. These are technologies that humans have developed and we developed them with our own strengths and weaknesses wrapped up in them.”
You’ve spoken before about this tendency of technology to unconsciously reinforce stereotypes. How do we prevent this?
“Well perhaps one of the first steps is awareness. That’s often an empty goal, because awareness without action is nothing.
“I’ve been involved in the heart of the technologist scene, as it were, for a long time and I’ve seen that the majority of the people involved don’t believe that they are building technologies that are biased. We really need to lift the debate so that we can talk about the various ways in which unconscious biases, and indeed conscious biases, are replicated.
“If anything, I think the current debate about Facebook and Twitter perpetuating hate speech is one of the strongest messages that we can get. It’s starting to shift a bit of the tone, from the Silicon Valley attitude of ‘we just build it, and it’s not our fault’ to thinking, actually, what we’ve built is something that we need to be responsible for.
“The problem with trying to change technologies that are widespread in the world is very clearly laid out. If it is a public company, then you have a legal obligation to report to the stockholders who have a very short term interest based on financial returns on a quarterly basis. If you said ‘let’s reduce our advertising revenue so we can reshape who we answer to’, that’s not going to fly. Whereas companies that are not yet public, or companies potentially like Facebook or Google who have majority control by the founders, maybe they can still go against that shortsighted direction.
“These are not so much problems with technologies but problems with the financial structures that support them. And that’s a lot to change. I do think that maybe it’s a problem of scale – we can create really good technologies that are fantastic on a small scale but that become inherently problematic once they become large scale. That said, one of the one of the things I believe is that if we can invent the problem, we can invent the solution.”
What are your biggest fears about the future of robotics?
“I think that we won’t end up with the really dysfunctional scenarios that people like to get scared about, like AI taking over the world or robots hurting us. Those things won’t happen, but I think the stupid short term problems definitely will. We’ve seen it happen where AI creates feedback loops that work on stupid things.
“We’ve seen financial trading for example, that causes billion dollar losses rather than smaller, individual losses. The feedback loop gets exaggerated and the trades happen faster with automation and we end up losing billions of dollars. We’ll see more of that kind of augmentation of human stupidity because AI and robotics allow us to augment human ability, and human ability can be stupid.
“We can also augment human bias. It’s in the field of gender that I see the most obvious problems happening. For example, an artificial intelligence like Microsoft Tay can get trained to hate speech.
“We’ve seen in video games where the representation of women is highly objectified, sexualised and stereotyped. Well I think with robots, because we won’t want to see such sexualisation, it’s going to be a little more subtle, but we’ve already seen that the female-looking robots and female-voiced robots are being used for the most menial of tasks. They’re very supportive, and it’s very difficult at a time of #MeToo, at a time where women are standing up and saying ‘I’ve just discovered that I’m not being paid as much as my males peers and I’m supposed to be at the top of my field so I’m going to fight’ and yet we’re building technologies to pass on the message that the woman’s role is not to be assertive, but to be compliant.”
Can you tell us a bit more about this notion of augmented stupidity and bias?
“One example is of AI being used to ascertain benefit status for health benefits. There is potential during construction of any system for conflict of goals i.e. best benefits vs lowest costs. The hardest thing about augmented stupidity is to find out who is at fault.
“We also have a problem with transparency in AI and in automation generally. Developers blame designers and designers blame commissioners, who blame the rollout and then everybody blames the person on the other end and says ‘well, they probably weren’t eligible for health care’. But its been found that people were being denied health care that they were legally clearly eligible for by systems that were skipping a few steps. Nobody took responsibility for checking this or working out what steps were being skipped.
“This is one of the most obvious examples of the power imbalance perpetuating augmented stupidity. The onus is put on the individual to complain and say they don’t think this result is correct, and no one is auditing it.
“To me, the two underlying problems with augmented stupidity – and the fact that it is, undoubtedly, rolling out everywhere – are that we are not putting effort into auditing and we’re not building and using transparent modules.
“Much of our AI and robotics is being done using off the shelf components. We’re using computer vision libraries that are not developed in house, we’re using software bridges that are maybe open source and maybe commercial but they’re not developed in house, we’re using navigation packages, mobile bases, sensors. There are a lot of things going into these systems and lot of trust that the system is going to do what we think it’s going to do.”
What are you most excited for in the future of robotics?
“Well founders, on the whole, are not really engaged with the idea of building something that is going to be auditable, accountable and make the world a better place. But right now there is a certain sort of founder who does congregate around the newest technologies of the world and is driven by the idea of doing things to make the world a better place and who will persevere in spite of of lack of funding and in spite of technical obstacles. I am privileged to be surrounded by people like that.
“That makes me very optimistic and hopeful about the future. And then I look a few steps down the pathway and I think our follow up is to talk about developing good design guidelines and helping governments talk about developing requirements for accountability.
“We are starting to be accountable for the diversity of the workforce. I believe having a diverse workforce would inherently shift some of these problems. First, you’re developing the accountability framework and second, we don’t know if we’re maxing out all of the potential solutions that are available with the homogenous core group that we have in technology right now. If we have have a greater diversity in that group, we might have many more solutions on the table.”
What companies are you excited about at the moment?
“Willow Garage was an amazing open source research and development group in Silicon Valley. It was founded by a wealthy individual who was involved with Google and sold his own companies as well. He wanted to build the first personal household robot, and it turns out, of course, that’s a really big problem to solve! So instead, he developed a framework – the first robotics platforms that could be used by multiple Universities because, at that point, robotics research did not have any standardisation.
“At Willow Garage they created the open source Robot Operating System which is now used by pretty much 100% of robot startups. They also developed the the best experimental robot, the PR2, and also the lowest cost open source robotics hardware platform, Turtlebot.
“But the key thing is that Willow Garage was a magnet for all of the talented robotics people in the world. It was fascinating because it was not a research centre, as such. While it was doing research and open publishing it, the goal was to develop things that were for commercialisation. While Willow Garage itself folded 2 or 3 years ago, it spun off 8 or 9 companies, and beyond that, almost all of the exciting robotics companies being developed in the Valley have people that worked at Willow Garage at the heart of them.
“Willow Garage has been a bit like Fairchild. It’s well known how 80% of the technology companies in Silicon Valley are Fairchildren, and that includes the venture capital firms as well. Willow Garage has had a similar impact, and if I look for an exciting robotics company, it’s almost 100% likely that someone from Willow Garage will be involved.
“One company I’m excited about is Mayfield Robotics who developed Kuri which is just being delivering to houses this year, having launched at CES in 2017. I have one in my living room right now! There’s a really fast turnaround these days for robotics.
“There are people who may not be the founders but are a part of the core team who worked at Willow Garage, and for me, this is a really good indicator that they come from the coolest group of roboticists out there – roboticists that have a sense of purpose as well, to make the world a better place.”
How will robots integrate with our day to day lives over the next 5-10 years?
“I believe that robots train us just as much as we train robots, we work together, it’s a very collaborative process. Anybody who’s used voice commands or touch commands in their car, or on their tablets and phones will recognise that.
“If we go into just about every bathroom (and I’ve been travelling a lot, so I’ve recognised that this is definitely worldwide), we expect to wave our hands to make things happen. I think it’s fascinating that the bathroom is the space in which this is happening. We don’t really talk about it much, but we’re training ourselves in haptic interfaces.
“I think over the next 10 years, people are going to use haptic, gesture and voice commands and expect things to happen. For me, the question is, are we going to get that right or are we going to realise that there has to be a better way to control our devices and start to develop a new control technology?
“For example, if we look down the track we might see something like an embedded or semi-embedded personal mic where we can tongue click our way through a menu. Or something that responds to our eyes, like AR or smart glasses or potentially in-home cameras that detect eye control with devices. Or perhaps we end up with a combination of the three things where you use a voice trigger or something to open a menu and then make selections from it by pointing and eye tracking where you point – the permutations go on!
“I’m going to be really interested to see what interface technologies we develop over the next 10 years. Everything in our house can be part of a robot house, and I expect that to include at least one mobile device, call it our home robot. I’m hoping that we actually have a couple of robot arms in household use by then. I have seen people working on robots to do things like dishwashing and toilet cleaning and cooking and folding clothes. But it’s always an open argument; is it better for us to build an appliance that is a one stop shop for cleaning and folding and hanging our clothes, or should we continue to have the appliances we have and have some form of robot arm that takes things out and folds them or puts them away? What’s the most effective? We just don’t know.”
If you’d like to collaborate on your next tech project, get in touch with Simpleweb today.