Robots risk perpetuating damaging gender stereotypes, Silicon Valley's robotics industry group's managing director has warned.

"There is a lot of stereotyping starting to emerge in robots. When all kiosk robots are female and all guard robots male, we are patterning our stereotypes and our biases and perpetuating them," Silicon Valley Robotics MD Andra Keay said in an interview with Techworld at Web Summit.

andra keay silicon valley robotics youtube web summit
Andra Keay © YouTube/Web Summit

Embedding gender stereotypes within robots could make it even harder to create more equal representation across different roles, and once created it is hard to reverse this trend, she warned.

"I remember speaking to a male teacher when my kids were in primary. It seemed hard wanting to do that job as a man. People looked at him as if he was strange. He was definitely the minority because almost everyone else was a woman. He was facing what women or people of colour face all the time," Keay added.

Part of the problem within robotics is that it tends to be quite a "demographically homogeneous discipline", as Keay put it.

"I'd like to throw a challenge back to the industry. I'd like to see more teams where people have more real-world experience beyond robotics. I see a lot of robots that fetch beer. Maybe there are other problems we can be solving but too many people in robotics are only exposed to a fraction of the problems in the world we need solutions too," she said.

However, elsewhere within the field of robotics, she painted a far rosier picture.

Investment and innovation

The level of investment in robotics has increased significantly from $1 billion in 2015 to $1.65 billion so far this year alone, according to Keay. By way of comparison, on-demand food delivery alone attracted $1 billion funding in 2014.

"That meant all of robotics as a field was getting the same funding as food delivery," Keay said.

"Five years ago I was cold calling investors trying to get them to come to our funding days. Now they are calling us saying they have identified robotics as a good area to invest in," she added.

This increasing level of investment is also pushing robotics into a whole new range of industries and applications, such as food production and textiles.

Growing trends include the development of nanorobotics, software robotics, microbotics, and 'soft robotics', which allows robots to handle fragile material that mechanical systems normally can't grip or would damage.

"Soft robotics is a lot cheaper. It is hydraulic or pneumatic where gas provides the structure. It's modelled on octopuses and how they operate: if you imagine a glove where when deflated it's floppy but when air added it becomes more rigid," Keay explained.

As a result of burgeoning innovation, decreased costs of the underlying technology and increased access to funding, we can expect to see the adoption of robotics accelerate rapidly in the coming years, according to Keay.

"Robotics is not so much a technical as a business problem. There's a lot of good tech but it's underutilised and under-required. So we're focusing on making better connections between entrepreneurs and robotics teams," she said.

Robots: job creators?

While software automation is undeniably starting to threaten potentially sizeable job losses, physical robots are not, according to Keay.

In fact, "what we see in the last five years of studies is that when firms introduce robots, they generally make the firm more economically viable, so they expand and increase their employment", she claimed.

"The most things robotics do is not replacing a job, but a task that is a subset of a job. And it's often the tasks which make some jobs unpleasant, which makes it harder to attract people to do them," she added.

In fact, Keay goes as far as to say we have an ethical imperative to introduce automation and robotics into areas where people do dangerous jobs, or are at higher risk of death or injury.

Agriculture is one of the best industries to look at automating and introducing robotics, according to Keay, as a means to address the issue of the need to rapidly increase food production to meet growing demand.

"Farmers are reaching retirement and it's becoming harder to get seasonal labour. There were two big increases in agricultural productivity in the last century: the introduction of the tractor, then chemicals like fertilisers. We haven't seen a productivity gain since then, but robotics and sensors could do that with precision agriculture," she said.

Robotics has an image problem

Unfortunately, in the popular imagination, robots are generally imagined to look or behave like humans.

This is a problem because they are the "smallest subset of robots in the world, plus the least reliable, least usable and least well developed", according to Keay. "We're just not that good at building humanoid robots yet," she added.

A good principle is to 'design downwards': in other words, only make robots as sophisticated as they have to be, building simple systems that aren't over-specified or surrounded by false expectations.

"I see a lot of companies trying to overhype and oversell. They're going to disappoint us and I hope it doesn't bounce back on the industry. There are so many good robots in development and while capability continues to improve, it's nowhere near a humanoid type level," Keay said.

Part of the reason we're yet to see humanoid robots is that the development of artificial intelligence plus facial, language, emotion and body language recognition are nowhere near good enough, she said.

"We need to get systems better at understanding this because easily 60 percent of our communications are non-verbal. Even verbal statements can depend on emphasis. Humans are very good at understanding this. But AI is really more artificially stupid than intelligent. It's not great at getting what we actually mean yet," Keay added.