Interview on Forbes website
Deidre Hughes has recently been interviewed by Michael Arthur from forbes.com. Full interview below:
Building A Bot To Beat The Bots: An Award-Winning Approach From Across The Pond
Deirdre Hughes has been making a difference on the UK career development scene for over thirty years. For the past 13 years she has managed her own company, DMH and Associates, to respond to a wide range of practitioner, employer, researcher and policymaker needs. One of the company’s claims to expertise involves using information communications technology to support individual career owners.That expertise is presently being employed in a major research project behind the title to this article. The project has developed a bot—an automated program that runs over the internet—that offers a refreshing take on how bots can enhance rather than weaken your career prospects.
Michael Arthur: It seems paradoxical to employ a bot to help fix the problems that bots are causing. How do you answer that?
Deirdre Hughes: We believe that bots and humans can work together. You hear a lot about bots and technology being disruptive, but they do not need to be. Our work is greatly influenced by the idea of social technology, of using human and digital resources to influence social processes. I have always believed that we have to harness technology, and to see it as something that will open up possibilities. Our work is all about empowering people, where a bot—our bot—gives you a personal agent, 24 hours a day, seven days a week, in the comfort of your own home. You can also enjoy backup support from local careers or employability advisors.
However, this means we also need to have a safety net for more in-depth conversations, to help people think through and act on their particular career situations. We also have to be sure our bot provides support to people who most need it. We call her Cici by the way, which stands for Career Inspiration Chat Information.
Arthur: So, the trick is to build a bot that recognizes when it’s out of its depth?
Hughes: Absolutely, and that’s one of the things we feel strongly about. We live in a world where algorithms can be good or bad. Often, IT experts bring all their technical knowledge and know how, and then they deliver what they’ve done to the practitioners. Here in the UK, policymakers want to think there’s a magic algorithm that will perfectly match individuals to a set of circumstances. Whereas we are reversing that logic. We started with the seed of an idea and worked with the practitioners—careers specialists with “boots on the ground”—first of all.
So, for us, the algorithm in Cici is about the conversation flow. And that conversation flow has to mirror the questions that you would ask every day when you’re thinking about a job change, or looking for careers information. You can have natural language processing based on the artificial intelligence embedded in a bot. Without getting into details, it draws on a combination of AI and social science approaches to engage in a conversation flow. And what I find exciting about our work is that you can train the bot to know the limits of it’s worth.
Arthur: Help me with that.
Hughes: You need to think about issues like “How do you avoid bias?” “What would the personality of the bot be?” “What accent would it speak in?” and so on. However, it’s especially important to anticipate the range of topics and ideas a user will want to explore, which means bringing practitioners on the journey. We have used practitioners from three different cities—Bristol, Derby and Newcastle—to share ideas about how they could use the bot. Working with those practitioners, we took particular account of the conversations that they had with their clients, and the very limited amount of time they had to spend with each of those clients.
You suggested it was paradoxical, but we have found Cici can help people to explore a wide range of options specific to their needs, not through someone else but on their own terms. We started the journey thinking she should be designed to help adults who were at risk of job loss from automation. However, we have learned from our testing that adults from a broad range of circumstances find Cici helpful. They can compare information about occupations, or identify careers and ideas that they haven’t thought of before. Moreover, as long as you are drawing on trustworthy information you can empower people to make good decisions.
Arthur: Can you say more about trustworthy information?
Hughes: You want a typical working person, the man or woman on the street, to be able to go in and actually search on different occupations. Cici knows where to go to draw on a range of data from various sources. For example, there are two government publications called the Skills and Productivity Report and the Labor Force Survey. Our bot’s Applications Program Interface (API) can extract relevant information from those publications and present it in narrative form to the user. Some university course descriptions may be indigestible to an average person. We can reduce mismatches and save people from going round in circles looking for information.
Arthur: With the advent of Covid-19, it has been demonstrated that a wide range of jobs can be done remotely. What does it mean for your focus on cities, if people can beam into those cities from outside?
Hughes: That’s a really good point, and Covid has already displaced some of the workers we’ve been focused on. My answer is in two parts. First, you really need the national and regional information about labor market trends and remote working to be trustworthy, so that users can see those trends for themselves. The second part of my answer is that local practitioners see a lot, and labor market information is one of the things that they can see. It’s tacit knowledge that lives within the practitioners, and it can be made available through them for the bot to share. So, I like the idea that Cici can be empowering for both career owners and career practitioners.
Arthur: Who are your competitors?
Hughes: We have made a firm bet on conversation flows, and on keeping conversations real. We do not promote combining AI with psychological testing, which deflects the user away from their own way of seeing the world. We have read about experiments going on in France, Israel and Spain, but none of them brings the same emphasis as we do to promoting a conversation on the career owner’s terms, and staying ahead of the game by enriching that conversation. We also did our homework on the US, but did not come up with much.
I always ask what I call a “wicked question” about value added. Professional bodies and politicians and practitioners have always got to ask themselves the question “If I was laid off tomorrow, where would I go for career support?” It is a hard question to answer. But if we cannot answer that question with some degree of confidence, then there is something wrong in the system, and in our society.
Arthur: A final thought?
Deirdre Hughes: Imagine what we could do if we got the architecture right, and got the trustworthy information in there, and then kept the conversation flows going. Imagine using our bot’s capacity to carry as many APIs as we would like, and switching them on and off to better support the conversation flows. Imagine Cici as a skilled helper that learns from her own experience and is always informed by regional practitioners. Imagine what a wonderful application in social technology that could be.
Michael Arthur: That’s something worth imagining. Thank you very much.
Work on the bot described here is being supported by a $70,000 grant from the independent charity NESTA, formerly the UK National Endowment for Science, Technology and the Arts.
The co-founders of the Cici team are Career development specialist Deirdre Hughes, researcher in technology development Graham Atwell, labor economist and career counselor Chris Percy and bot developer George Bekiaridis.