Dehyping Robotics and Artificial Intelligence (AI)
Dr. Sabine Hauert, President and Co-Founder of Robohub.org and Assistant Professor in Robotics at the University of Bristol, provided the InterConnect 2017 audience with an insightful (and interactive) discussion about robotics – highlighting the need for balanced media and communications around robotics and artificial intelligence.
As a member of the Royal Society’s Working Group on Machine Learning, Dr. Hauert is an expert in science communication and a frequent speaker on the future of robotics. In her talk, Hauert explains how robots can be game changers, but not in the way we think.
How and where can robots assist people?
Robots are not going to replace humans, they are going to make their jobs much more humane. Difficult, demeaning, demanding, dangerous, dull – these are the jobs robots will be taking. Productivity is one of the primary benefits of robotics in the workplace. In Europe, the goal is to attain a 20 percent increase in productivity by 2020. Central to achieving this is the exploration and use of robotics in the workplace.
Fact: There are 1.25 million deaths caused by automobile accidents every year. We spend approximately 200 hours a year behind the wheel of an automobile. Autonomous cars could give us back this time, while being safer and greener. Autonomous vehicles can also be empowering for millions of people who can’t drive today.
Health and well being
Fact: One in four peoplewill be disabled by the time they reach retirement. Robotics, prosthetics and exoskeletons can provide technology that can restore freedom to people who now experience limitations due to age or disability. But, helping disabled people through technology is not about removing the human element.
Hauert asks the audience for a show of hands: ‘How many of you would want a robot to take care of you as you grow old?’ For most people, the answer is no. She then asks, ‘How many people would be okay with a robot helping them to get out of a chair, get into bed, go to the toilet, or fetch a glass of water?’ Hmm – a lot more of the audience say, yes.
Elderly care and aging is a global issue we are all going to experience. We are going to need help. Robotics is one solution which can be used to help alleviate these challenges. If robots can help individuals with the day to day tasks, this leaves more time for doctors, for family, for care takers to participate in other things they care about, and which they are more competent – the discussions, the conversations, the human contact, the act of caring. In this setting, the robots enable the elderly or disabled to enjoy the things they want to do – in either their own homes or assisted care facilities.
Empowering people with disability
The use of technology in humans is not about creating super humans, it’s about empowering people with disabilities through assistive technology. One example is the Cybathlon, an international competition organised by ETH Zurich for disabled competitors allowed to use bionic assistive technology, such as robotic prostheses, brain-computer interfaces and powered exoskeletons. The Cybathlon is much more about human achievement than the use of technology for the disabled.
The rise of the cobots
Take the example of a Baxter robot – or a cobot. A cobot is a collaborative robot designed to work alongside a human. Cobots are designed to be taught to do specific tasks within a company. By taking its arm, and showing it what to do, the cobot puts its human collaborator in a position to teach it skills, to show it how to perform functions, to manage its activities.
It’s an intuitive way of approaching robotics in the workplace which puts the human worker in a position of responsibility to improve the productivity of that company. You don’t have to be an expert to program a cobot – but you can become a manager by teaching cobots how to assist in specific tasks.
Precision farming and agriculture
How is it going to be possible to feed 10 billion people by 2050? Precision robotics used in farming is one way to solving this challenge. Food production and supply is another example where robots can provide valuable assistance.
To a certain extent, precision farming enables farmers to look at how they approach crops in a more personal way – by understanding water and irrigation issues better, by using data from soil – for moisture and nutrients, to understand more accurately how to pinpoint the use of fertilizers and soil additives, or to reduce pesticide use.
Exploration of the universe
Fact: The most prominent robots in the world are used to milk cows. Granted, dairy farming is not the first thing that comes to mind when most people think about robotics. A milking machine is not as sexy as a robot designed to help us explore our universe.
Exploration and discovery of the whole universe – deep sea and outer space is another area where robots are helping humans to conduct research, and explore dangerous or inhospitable environments. Robots offer exciting new ways to explore the earth – in mines and caves, in the deep sea, the air and even outer space.
Designing swarms that work in large numbers, and at small scales
Swarm robotics is a new approach to coordinate the behaviors of a large number of relatively simple robots in a decentralised manner. Taking inspiration from complex behaviors like swarming can help to design technology. Before joining the University of Bristol, Dr. Hauert engineered swarms of nanoparticles for cancer treatment at MIT, and deployed swarms of flying robots at EPFL.
Hauert’s research at the Bristol Robotics Laboratory focuses on understanding and applying the principles of swarming to robotics. By studying starlings which can create beautiful and complex patterns, Dr. Hauert is able to use the principles of swarming to better engineer robots that work in large numbers. Hauert started out designing swarms of robots that could be used to potentially create communication networks in disaster areas. In Bristol, Hauert’s team explores how to design algorithms that work with thousands of robots to potentially search an environment.
Why does the media insist on portraying robots as scary?
A member of the audience asks the question, ‘Is there a better way to manage the impact of the story if the press is determined to use a clickbait headline?’ Hauert recounts a story where a news outlet claimed her intention was to inject millions of nanoparticles into people.
Hauert jokes, ‘Roboticists are not evil. I am not an evil person.’ But I am interested in nanoparticles. Nanoparticles number in the trillions. Hauert is interested in the use of nanoparticles to treat tumours – specifically how we can design these particles so they work more efficiently, so they behave the way we want them to behave in a collective so they are more efficient when treating cancer patients?
Every green dot in this picture is nanoparticle under a microscope. This work is what Hauert perceives as being incredibly beneficial to society. Yet, the media doesn’t always portray the work she’s doing in the same way. In this instance, Hauert had to negotiate with the reporter covering the story in order to change the headline. The truth behind the statement lay in Hauert’s interest in and research on the use of nanoparticles in cancer patients. By injecting nanoparticles into cancer tumours, she was working towards a cure.
Another headline flashes up – ROBOTS to SLICE YOU OPEN AND CUT YOU UP
Are we astonished to learn that Google plans to slice humans up with robots? The mismatch between the headline and the subject of the article is startling – as the article is really about a surgery robot – something which, if developed, could be extremely helpful to improving surgical techniques and recovery times. But, is the true nature of the story reflected in the headline?
Does this happen in other fields? When surgeons develop new techniques are there headlines which say, ‘Dr. Evil wants to hack off body parts?’ Probably not. Robotics, more than any other field, seems to be tarnished with the brush of ‘science fiction convergence’ where everything gets related back to the world of science fiction.
Scary headlines are clickbait
Headlines are misleading – and don’t do the field of robotics any favors. Many statements and headlines are just click-bait. No doubt, the scary headlines receive more views, but what impact does the use of such inflated statements have on public opinion. Are these types of headlines just perpetuating the stereotypes through misinformation?
To help neutralize their effect, Dr. Hauert suggests roboticists need to actively engage the media. It is the responsibility of the roboticist community to promote better communication with the media by reaching out more frequently with materials that offer balance. While the roboticist can’t change the behavior of people who are drawn to ‘car crash headlines,’ they can work towards educating the authors of the materials to better inform the public.
Empowering experts to become better communicators
Robohub.org is a platform that connects the robotics community to the world. Hauert is on a mission to ensure balanced, truthful and fair robotics information is being communicated on a broad basis. The key to dehyping how robotics is portrayed – especially through mass media outlets – is to empower experts to become better communicators for their work.
Scientists and field practitioners need to learn how to communicate with a broader set of stakeholders. Rather than communicate in a bubble, Robohub.org is set up as a non-profit that connects roboticists with the public – using tools like blogging, social media, videos, and similar channels go gain access to a broad audience.
Moving roboticists outside of their bubble and into the mainstream is the central philosophy behind Robohub.org’s mission to de-hype the mainstream perception about robotics and artificial intelligence.
Will the use of robotics and artificial intelligence lead to jobs being displaced?
Once again, common headlines often paint a dim future where robots will take our jobs and destroy the universe.
Hauert cautions: ‘Before setting out on a discussion about job displacement, the full picture needs to be explored. Yes, there will be jobs which are displaced. However, there will also be new jobs created.’ A better question to ask is, ‘What kinds of jobs are likely to be replaced?’ More than likely the dirty, dull, and dangerous jobs, jobs which are better suited for industrial robotics are likely to be displaced.
Productivity improvements lie at the heart of robotics adoption
Understanding the reason why robots are being used in industry is a good place to start. Improving productivity levels is a priority for all organizations. Efficiency gains are driving the use of robotics in industrial environments, factories, routine service environments and manufacturing plants. Using robotics in industrial settings improves productivity.
In most instances, an increase in productivity leads to an increase in potential jobs being created within a company – either to fulfil new roles and tasks required to train and manage a robotic process, or, freeing up resource to deliver services or tasks which require a different level of human interaction.
When it comes to assessing the long term economic impact of robotics adoption and use, the data is not yet reliable, it’s just now being created. We are at the beginning of the cycle – in Europe, for example, there are no benchmarks yet. The clock has just started.
Is it helpful to put a face on a robot?
It depends – Hauert is divided on this question. Admittedly, looking human is a complex process – expressions are hard to imitate. For the most part, industrial robots are not human-like. However, in settings where robots need to interact with individuals, having human-like expressions or facial features can be helpful. Take the example of a robot designed to help autistic children – a robot that can display facial expressions is more likely to connect with the user on a human level. Conversely, a robotic device designed to hoover carpets does not need facial features, as it’s expressions have very little impact on the task the device is equipped to complete.
Is there any advice for start-ups interested in robotics?
It’s a lot harder to design hardware than software. Anything with moving parts and a body – no matter what shape, requires much more research and work than the software side. Hauert emphasizes the growing community of robotics professionals. There is a robust ecosystem of partners, scientists, think-tanks, legal groups available to provide assistance to budding robotics engineers, designers and entrepreneurs.
To help robotics start-ups, there are many incubators located all over the world which can assist small organizations with coproduction and scaling up. Access to communities of researchers and open-source collaborators, in addition to crowd-sourcing platforms, can also help to make the process less daunting. From a financial perspective, the use of crowdfunding platforms like Indiegogo and Kickstarter can help to connect investors and backers with entrepreneurs.
Hauert is adamant when she points out the importance of utility. Hauert emphasizes the need to design real products. Prototypes often miss the mark when it comes to designing a ‘product’ with a differentiating and useful set of features that make it stand out as a real product. Start-ups should create a product that is also a robot instead of a robot that is not a product.
Taking the hype out of the reality
What is the prevailing opinion when it comes to robitics? During the session, Hauert discusses findings from research, studies and interactions which reveal the public’s real perception of AI, machine learning and robotics.
Fact: Only 9 % of people know what machine learning is. Yet the majority of people do understand the concept of Natural Language Processing (NLP) and autonomous cars.
How do people know about NLP and autonomous cars? Seventy-five percent (75%) have learned about these concepts through mainstream media, while another twenty-one percent (21%) attribute their knowledge to entertainment – television, films, or books.
Through further investigation, the findings also reveal we have a healthy interest in the direction and use of artificial intelligence, robotics and machine learning. For example, ‘Can we trust algorithms to work with humans in a safe way?’
And, yes, people do worry about robots replacing jobs, and they express concern about the possibility that robots might replace the human side of caring services.
Other interesting points, the public doesn’t really know if machine learning is good, or bad, but they do have, and are voicing more contextual concerns relating to data – for instance, ‘Is there any evidence to suggest that autonomous cars can improve safety records?’
Lastly, people care about productivity dividends. If productivity increases, many people believe the result will be knock on benefits of wealth down the line. A common concern expressed is whether there is a way to redistribute the bi-product of increased productivity – wealth – in a way where everyone along the chain can benefit.
Here’s how you can help to de-hype robotics
Dr. Hauert closes the discussion by asking us to remember four things:
1, Making robots and AI is difficult – it takes years to design and perfect.
Take the example of an elderly care robot. In order to complete a task like get a glass of water, there are hundreds of actions, algorithms, hours or research in order to get a robot to complete any particular task.
2. Robots perform tasks not jobs
Highly specialised AIs can complete very specific tasks – such as vacuuming a room. This is a far cry from The Jetson’s Rosie the Robot who cleans, cooks, answers the door and walks the dog.
3. The jobs of today will not be the jobs of tomorrow
The job market evolves constantly relative to technology. The use of remote working, video enabled conference calling, digitally enhanced learning environments. The jobs from the 1950s are not all relevant in 2017. The same principle applies to the jobs of today – they will simply not be the same jobs available to the workforce in twenty years’ time.
4. Robots are different – software bots and hardware bots are different
The process is different – one is a device and the other is a piece of software. While the first might require the second, the second does not need a hardware shell to function. It will take a lot longer to design a fully functioning robot that will be able to understand language, apply reason and move in intricate and delicate ways – all in the same ‘cyber’ body.
For the public and anyone involved in robotics, it’s important to continue to ask questions and to voice concerns, as well as aspirations. What regulation needs to be put in place? What ethical guidelines need to be put in place? How can we empower start-ups? Are there legal frameworks to assist us in setting boundaries? What can we do to ensure the benefits gained through the use of robotics are available to everyone?
The more questions that are asked, the more granular the answers become. In the meantime, please connect with the robotics community. You can find Robohub.org online and connect with Dr. Sabine Hauert on Twitter.