Posted by Since April, I have applied for two lecturing jobs at Russell Group universities which appeared on jobs.ac.uk.
My main reason is that since setting up the AI Co-Creators, having the backing of an academic institution is the easiest way to get some funding to carry on doing what I do. I look at the jobs board anyway just as it’s a quick and easy way to see where the funding is going and as someone who reviews for UKRI (I was elected to the EPSRC Peer Review College in 2002) from outside of academia, it is a shortcut to what departments are focusing on and what funding they are currently chasing.
There were two jobs for which I am suited and would get me leg on the table in an academic institute.
The first was teaching AI online and as I do teach AI online already, it seemed like a good fit. However, it definitely didn’t feel right when I asked, during an in-person interview panel, how many students per cohort? I was told up to 100. Multiplying that by two courses and three academic terms and the second-marking load – the answer I got when I asked about teaching support (aka none) – gave me a conservative estimate of 900 scripts a year. I do not see how I would have time for research at all even if it was theoretically ring-fenced a day per week. I was exhausted just thinking about it. I was also told that they had AI experts already in the department. Well what did they need me for?
Luckily, I had already gone around the panel asking everybody what they were learning on Duolingo which makes me laugh even now. It wasn’t planned, it was just for me, a useful exercise demonstrating consistency in online education. How much these people squirmed at admitting that they weren’t very good at consistency and showing up each day, while seeming to think 900 students, half their age or less, would magically do so without any support from them. One panel member refused to join in at all saying they didn’t do anything. But since they were the one who had asked me how I would teach decision theory and I had to give a polite answer rather than saying I don’t plan on time travelling back to the ’90s anytime soon, I don’t think I missed too much from not hearing their thoughts on Duolingo. All in all it wasn’t a good fit for me or my research or my joy. I did follow up asking if they would consider hosting me for funding – I am still waiting for their answer.
In July, a position popped up and I wasn’t certain but after thinking for a while, and rereading it, I recognised myself in the job specification. It was for human-centred AI (HCAI). Something I have been doing since I moved to Switzerland in 1995 to begin doing research in both HCI and AI. The more I looked at it, the more it seemed exactly me. I present a little bit differently to a typical candidate so I thought about the job, wondering whether to apply or not as it was perfect for my research. It would allow me to do all the things I do anyway. And also I just love students. In July, an AI MSc student came running down the street after a talk I had just given in the pub to tell me she wished she had lecturers like me. I make AI really interesting and explain it clearly.
Inspired by her enthusiasm about my teaching style, I figured well I have a decade of university teaching (HCI, web technologies, Cognitive Science, AI) under my belt. I have an online HCI course which 1,323 students took completely voluntarily. I review research for various institutions. I have given 20+ public talks about AI over the last couple of years and I have lately been engaging in some research that could go straight for funding now. Oh and this blog – nearly 20 years of writing about HCI and AI with respect to our digital culture getting 1.2 million hits a year -referenced and used in academic teaching by me and others across the world. In a worst case scenario I would learn something from the process as applying for a job and all the hoops I must performatively jump through gets my juices flowing. I not only produce something useful, I also get to look objectively at myself which honestly, just dazzles me.
I clicked on the apply link and was just taken to a webform with name, address, etc., and then further down, or page 2 on my phone, it looked like this:
This looked a bit weird and what I wasn’t expecting. I was expecting to hand over a CV, a research plan and a full supporting statement which involves cutting and pasting all the headings from the job advertisement and answering each statement even sometimes repeating myself so that the HR people can go through find the words, match it the the job advertisement headings and tick their boxes.
I hadn’t ever seen ‘Research Paper 1, Research Paper 2’ on an application form before, it seemed weird. Undaunted I looked at my publications page to find two suitable journal papers that I have and thought yes I would upload them. My academic publications are older, and blogs don’t really count — though they say they do, and mine are enough to make any ‘serious’ academic faint – nearly 20 years and half a million words of pure me.
So, I went off looked for a call for papers that was upcoming, wrote a short paper based on a blog-post about the research the AI Co-Creators and I had just done: Shaped by the Learner and submitted it to Interacting with Computers. It was accepted with conditions and so I can’t wait to write that up. I even wrote another blog to pull together some thoughts so that they are written somewhere as a holding page, before I get around to the whole paper. And now I’m on a roll. I have quite a few papers I want to write up from my research there and another longer-term project involving the coding up of a SLM which is very exciting.
A couple of days later, in a make-work scenario if ever there was one, I was forced to complete the ‘qualification certificates’ box by faffing on scanning my certificates (PhD in HCI and AI, BSc in Computing, cycling proficiency — not really, but I did put in PGDips and equivalents: teaching, AI, journalism, VR, etc., etc.) and then I cut and pasted my supporting statement into a box here as there was nowhere to put it, not a single ‘supporting statement’ box in sight:
By playing with the boxes by cutting and pasting in text, I found that new boxes would present themselves in a responsive design approach, which means that the form will update itself when I add enough text to overfill that box and then generate another box for me to add in more text. I could do a maximum of three boxes of 4000 chars so it was a bit tight but I did it. I got my supporting statement in there. Really in this scenario, progressive disclosure not responsive design is what everyone should be doing – which is when the user completing the form can see there are links to opening up another dialog box of more boxes if you need them or you can just upload an attachment – you don’t have to turn full HCI expert just to investigate, understand and complete the application form.
Completely frazzled, I submitted my application.
An hour later a human emailed me directly and said that they wanted me to withdraw my application and gave me a specific list of things they wanted to see: CV, supporting statement, teaching experience, and a research plan.
This is known as Invisible Infrastructure, well the precise definition is slightly different, either way it is well-documented in the HCI literature that the systems are so broken, so incomprehensible, you have to have a human going about telling people what to do as the system cannot. Why? Because the system is terribly designed and not serving any purpose. Research paper 1, research paper 2 indeed. It looks like some 14-year-old was asked to design a webform for university applications.
At first, I wrote back saying I had already done all that but they never got back to me. So, after realising that they weren’t going to reply, I took another look assuming they would have updated the web form to reflect what they wanted. They hadn’t.
So! I had to get creative and upload in Research Paper box 1 and 2 all the things they needed concatenated together so that it would present correctly. If I had written the guidance email, I would have written a list in the actual email telling which box they wanted things in e.g., Research plan + teaching plan in Research Paper 1. Statement of support in Research Paper 2., etc., to make sure I was being clear about my requirements.
I faffed on some more, then resubmitted my full application, then realised I hadn’t withdrawn the one before, so deleted as much as I could see – this form part of another system was so bad I didn’t even take a screenshot – then resubmitted. Then I got an automatic message confirming that I had submitted my application. Then I emailed the human saying I had resubmitted according to their email specification. I didn’t want to look like I was criticising their systems so for me the subtext of my email was: if you can’t see exactly what you are expecting then email me. A couple of days later I got an email back from said human saying: Thank you, we are looking forward to processing your application.
A couple of weeks went by and I hadn’t heard anything even though the advert was still up and people were hitting my website -they went deep too all the way to the blogpost on my Wonder Woman Hotpants. But, the time between application deadline and interview was two weeks, though the job ad was up for nearly two months. So, on the day the application closed, I emailed the human saying that my paper had been accepted to see where we were and remind them that ‘Hey sweaty human here waiting to hear from you. It’s really important to me.’ I got a thumbs up emoji as a response.
Little did I know that would be the last I ever heard from them so it’s a thumbs down from me for Russell!
I wish they had at that point told me the outcome. Surely the shortlisting had already happened if candidates were expected to book time off and prepare for interviews. The teaching AI position panel had asked for a 20 min video about my teaching and research, which I did as a talking head (and someone I met on that day said my video stood out by a mile which was lovely of them but so inappropriate). I have so much experience now making videos and talking heads I am so much better than I was ten years ago.
I digress, the first round of HCAI interviews was conducted today and typing this tonight, I still haven’t heard anything. I guessed I didn’t make the list, but since no one had told me that I hadn’t, this evening I logged into the ‘portal’ and saw this:
It’s like I didn’t apply at all. My applications, my experiences, my qualifications, my whole academic and research life are no longer represented. They have completely disappeared from the system.
Now, I don’t actually know what happened to my application. It may have been archived, closed, rejected, or moved somewhere I can’t see. The system doesn’t tell me. The irony is that on their institute webpage they describe themselves as:
A world-leading centre for research and teaching on the connections between people, technology, and society drawing on the best scientific traditions in Computer Science and Human Sciences.
They can say whatever they want to on their web page but their behaviour communicates something else altogether about the institution itself.
System error or systemic disrespect?
I can’t speak for them, but when I think about infrastructure and processes involving humans, I want them to transparently map onto the webform and all other contacts in order to handle the human experience from start to finish. I think about designing dialogue and the other 101 things that make up a system based on solid HCI principles. I think about the processes of how a department works and how to support those processes and people using software, and how they are represented. I also have many years thinking about AI and how to get it to support humans (GenAI does not do this in the ways it could) and how it represents their knowledge. It is how I began my research.
Nowadays everyone thinks by putting AI on the top of template webforms and broken systems that somehow magically these problems will get sorted and the human will be looked after, which is why they wanted someone with a computing and AI background in the department to begin with. And, I know admin is not done by academics but it adds up to the one same user experience that I have of that institution. AI isn’t going to solve this mess, that is for certain and in this case where I have completely disappeared, there wouldn’t have been time for the AI to have registered I was ever there, it would be like I never existed. Perhaps that is more convenient – it is definitely less messy.
Any given system may seem to work well enough from the institution’s perspective while leaving the person, in this case me, with no idea what has happened. Something that is heavily documented in the HCI research literature beginning in the 1980s about systems flow and human behaviour. A human reached out to me soon enough when they wanted something, but I haven’t heard from them to let me know the outcome.
Disappointment aside, I’ve got work to do. Research that could never have been done in such an institution, which talks the talk but alas, doesn’t walk the walk. Though, I won’t lie I would have happily joined it to get the voices I represent added to the scholarly conversation and into the literature and our hands on the funding to build something that will make a difference.
However thanks to wrangling with the application, I have pulled together a whole map for my research program about AI-mediated representation, something I started blogging about here: Smoggie Queen (3): Stereotypes, Representation, and Lossy Compression in old and GenAI though basically I have been thinking about for the last 30 years which started with my PhD research and how to get AI to represent the people and knowledge who will use it.
Actually if I’m being honest, it started even earlier as a young woman and my place in society even before I had heard of representation in computing and AI, after all I was going to do English literature – and even then women didn’t come out too good. I didn’t build the research plan for the job, I had lots of different things going on and I pulled them together to present to a panel of academics. I was delighted to find that they added up to a great Gestalt whole all about representation, so that is what I learnt from the whole process.
And, representation is what my favourite talk, When women were computers, which I am giving tomorrow night in honour of the WI’s 111th Birthday, is all about, and it is the reason why I set up the AI Co-Creators.
I now have a whole schema for individual-community-institutional levels of representation which all went in my disappeared research plan for a job I may never hear about ever again but the experience of applying for this job quite tidily demonstrated in macro, the way I view the world and the levels of how we are not represented in and by IT systems which don’t work.
It is also feedback too in its own way, the AI words I used in my research were not the ones they had in their HR checklist which is probably worth a blog in itself but for now, leaves me with only one question:
Who’s going to fund my work?
Not to worry though, I am already working on the answer.
I can’t wait to see what happens next, not least of all how I am going to shoehorn my big funding proposal into a bloody Google Form! Because that’s how ‘world leading’ HCI and AI things work these days.
One final point I would like to make, but really I’m talking to myself, is this:
The question isn’t why they didn’t shortlist me. It’s whether the institution’s recruitment representation of Human-Centred AI is actually broad enough to represent the field it claims to recruit into.
Because if someone who has been doing HCI and AI since the 1990s can disappear from the representation, and be removed completely from the system like she never existed, and never applied, that’s not just a story about one application.
That’s much bigger than ‘just’ an HCI problem.
Something to ponder as I move further into my representation research.
Oh and use this as an example of UX design for my HCI course!
And finally, if they ever do get back to me to say I wasn’t shortlisted, I think I’ll reply with a thumbs up emoji and a link to this blogpost.