Operating at the Boundaries: Augmented Intelligence
“The desire to economise time and mental effort in arithmetical computations, and to eliminate human liability to error is probably as old as the science of arithmetic itself”
Howard Aiken, 1964
Augmented Intelligence, which we have previously written about, is the application of machines, humans and processes coming together to create powerful new insights. To help bring this topic to life, we invited a cross section of experts working in the fields of data science, philosophy and academia along with those applying it in the real world to speak at our fifth Operating at the Boundaries series event.
The series, run in partnership with The Royal Institution of Great Britain has previously covered high-performing teams, elite sport, storytelling and productivity.
Hosted by QuantumBlack’s Chief Operating Officer, Chris Wigley, our Augmented Intelligence event saw data scientist, Martha Imprialou — also from QuantumBlack — joined by Dr. Karina Vold, philosopher of mind and artificial intelligence researcher at Leverhulme Centre for the Future of Intelligence (part of theUniversity of Cambridge); and James Hewitt, Head of Science & Innovation at Hintsa Performance.
The speakers discussed and debated many topics around how Augmented Intelligence will affect our lives, the dangers of influence, how we learn to speak and work with machines, what the tools we use may look like, who to blame when a machine makes a mistakes, and what it will mean to be human.
You can watch the full video from the event below, or continue reading for a summary of the event.
According to Imprialou, technology augmenting human abilities is nothing new. From using a stone as a hammer, to using quantum computing to solve big problems facing the world, humans have turned to tools to aid them be stronger, smarter and more capable.
But in terms of more modern forms of augmentation between a human and a machine — in the context of Augmented Intelligence — Wigley provided examples of machine learning in action such as China’s use of machine learning in surveillance and Google Assistant making a human-sounding phone call. Each revealed how machine learning can be used to simplify tasks for humans or spell a potential risk, depending on how it is used.
All of the experts delivered both measured warnings about the need for those leading the development of artificial intelligence and machine learning to ensure protections are in place and that systems are not built with bias. However, there is hope and belief that Augmented Intelligence can enhance human ability with the support of machines.
For Imprialou, there are tasks which humans are better at, such as critical or creative thinking, and others which machines are better at, such as being able to take vast quantities of data and parsing it to find patterns. The trick is to identify which tasks each is better at and augment with either human, or machine.
For example, with prediction, which is the most common task machine learning solves — a human analyst takes a very long time to produce a forecast. A trained machine is able to perform the analysis and deliver output in a few hours, rather than several weeks, if not months. Yet, a machine can only rely on past information to predict what is coming. Humans have the ability to use intuition to deal with unforeseen circumstances or events better than a machine can. With this example, Augmented Intelligence comes to prominence by recognising where a machine can help offload work, allowing the human to think more creatively and analytically.
Finding a balance between tasks improved by human augmentation, or machine augmentation is important. But as the first fatal crash by a self-driving Uber car involving a pedestrian revealed, machines are only as good as how they are built. Other examples bring to light that no system is perfect. For example, Google’s sentiment analyser was shown to rate being a Jew or homosexual as negative, while Microsoft’s Tay Twitter bot was trained by Twitter users to have extremist views.
Systems that contain bias or are trained in the wrong way, or with bad data won’t deliver on the true value that Augmented Intelligence can have. As fellow QuantumBlack data scientist, Ines Marusic discussed at SXSW there is a need to remove bias and check fair algorithmic design principles are used.
This is especially true when considering the impact that technology can have on the brain. Vold made the case that care needs to be taken around privacy, data and manipulation when it comes to augmenting human skills with machines. As Vold explained, “The mind is fragile, according to the current thinking with cognitive science. And as technology has become more functionally integrated into our cognitive capacities, it’s increasingly on a par with the brain in terms of how we remember and how we’re influenced.”
Put simply: our minds are merging with technology.
Philosophers have long debated and argued about the impact that false information can have on the brain and how we think. History is littered with examples of how mass information and communication techniques have been used as forms of manipulation, propaganda and control.
In 1565, Swiss philologist Conrad Gessner warned about the effect of information overload, caused by the printing press leading to an overabundance of data being “confusing and harmful” to the mind. The same is now being debated with the impact of social media and mobile phones on how we think.
In many ways, this is an area that Hewitt delved into, albeit with a different take. As someone working in performance science, Hewitt shared how our brains can struggle to stay focused when we feel overloaded by working levels or distracted by technology. Augmented Intelligence has the potential to help remove distraction — when applied in the right way.
But today, it isn’t just the impact of how technology may be changing how we think — but that it also knows more about us than we sometimes know about ourselves.
The potential dangers of new advances that seek to merge technology with direct-brain interfacing was touched on by both Vold and Imprialou. The argument to protect data grows when thinking about artificial intelligence and machine learning — when applied to Augmented Intelligence — as it opens up greater potential for abuse and mistreatment when used in the wrong way.
Vold discussed the concept of the computational theory of mind, where thinking is a process of symbol manipulation. Taking this further, Vold shared how the latest thinking is that it’s not just the brain doing the thinking for us, but instead biological and artificial resources working together to explain our intelligent capacities.
As Elon Musk discussed at SXSW earlier this year, firms like Neuralink have a goal of exploring technology that will make direct connections between a human brain and a computer. However, with the increasing connection between our brains and the technology we use, there are concerns around advances with artificial intelligence.
We need to ensure that the right protection is in place for our cognitive capabilities. Vold argued that advertisers have long sought to influence our minds, with the aim of influencing or changing behaviour. As recent news headlines have highlighted, the aim of influencing or changing the views of a group of people continues to happen — social media has made it easier to share information, whether it’s true or not.
But it’s not just technology being able to read our minds, or being able to take action based on thought. There is also the trust that we have in the systems that we use. This is especially true in medical applications, where doctors and patients alike need to know that systems are able to identify the right course of treatment.
It also extends to systems such as autonomous vehicles, so that we know we won’t get killed when we get into the car. There has been a large amount of debate about how vehicles should make decisions about what should happen in an emergency, and who is ultimately responsible if a vehicle does cause damage or kill someone.
Technology already allows researchers to measure how our brains react to different stimulus and to see a live map of what is happening in 3D using a mobile EGG headset — even as we walk around. For Hewitt, the benefit of this technology is to help understand what optimal performance looks like and to understand how to get the most from humans, when paired with machines.
By understanding the human brain better, and the rhythms of attention, Hewitt highlighted the need for more of us to shift how we structure our day. As many of us can attest to, it is often when we are distracted or less focused on a specific task that it feels like we have a brain wave. Enabling the brain to have time to recharge is important; Hewitt urged us to switch off from the devices and noise that can cause cognitive overloading.
Hewitt believes that technology used in the right way can indeed help provide humans with the ability to think more and find creative solutions; rather than constantly moving from one task to another. Where Augmented Intelligence may have the best application is to help remove distraction from human thinking — providing us with the information that we need to help make informed decisions, without overwhelming us with distraction. As Hewitt put it, when it comes to F1 racing drivers — rather than focusing on the whole track, just concentrate on the next corner.
The skills we will need in the future will, and are already, changing. In the late 1980s and into the 1990s, companies had to invest huge amounts into training staff about how to use computers, and then in the late 1990s and early 2000s, how to use the Internet. More recently, companies have had to train staff on the right way to use tools such as cloud computing, social media and increasingly understanding what big data actually means.
Our colleagues at McKinsey Global Insights recently published a discussion paper discussing the skills that the workforce will need by 2030. The report found that basic cognitive skills, such as data input and processing will decline in demand by 15%. While demand for physical and manual skills will remain the largest category of workforce skills, it is also expected to have dropped by 14%. In terms of demand for skills that will be needed, technological skills will rise by 55%, in addition to demand for social and emotional skills such as leadership — which will rise by 25%.
Much of the change in demand for skills will come from changes in technology. With the next revolution in how we live and work being driven by advances in artificial intelligence and machine learning, it is these technologies which are simultaneously being hailed as either the end of the world as we know it, or the start of a new era of better productivity and job creation.
From driverless cars to voice assistants that make human sounding phone calls, artificial intelligence is advancing outside of laboratories in an increasing number of ways.
There are still a great many questions about what the true impact of Augmented Intelligence will be. The benefit we have now is that we’re able to use machine learning and artificial intelligence to augment how we find the solution.
The history of technology is the story of augmenting human limitations with machinery or tools that enable us to do more than our bodies or minds let us. But with every advance and change comes fear.
As the quote from Howard Aiken reveals, the pioneers of computers set out to create tools that would help augment human intelligence and ability with machines designed to drastically reduce the time it takes to complete tasks.
Ultimately, this goal remains as true today as it did when Aiken first discussed it. The key difference now is that we have access to powerful machines that are already having a proven impact on the speed at which big problems can be solved. We’re already using Augmented Intelligence to help people and companies be the best they can be.
For further viewing, here’s the Q&A from the event:
Further reading//
- Aiken Proposal, History Computer
- The Real AI: Augmented Intelligence, QuantumBlack
- Tackling Fairness: Gender, Diversity & Algorithmic, QuantumBlack
- Is Your Smartphone an Extension of Your Mind? By Karina Vold for Vice
- The Computational Theory of Mind, Stanford
- Opening Exercises: Thriving in the Information Age, Princeton
- Are ‘you’ just inside your skin or is your smartphone part of you? By Karina Vold on Aeon
- How artificial intelligence can deliver real value to companies, a McKinsey discussion
- Skill Shift: Automation and the future of the workforce, a McKinsey Global Insights discussion paper
