The New Developer research project from the Developer Success Lab + yours truly is now available as a preprint -- we hope this will facilitate easier downloading/sharing/and citing of the work
In this work we study how beliefs about brilliance & toxic "contest cultures" undermine software teams' capacity to confront the last few years of genAI in software e.g. by fomenting AI Skill Threat. But we also explore the protective effects of learning and belonging cultures
@grimalkina I find this cool, but I have strong problems with the way the paper seems to embrace the fact that learning to use AI is "upskilling" and something that we need to adapt to.
I am particularly surprised to see no support for this model in the paper. I understand where you come from. And I like a lot of stuff in there. But damn does it make it hard to adapt it to my reality.
@Di4na hmm that's not my stance at all and I will say I appreciate your concern but I think it's a misread of the paper -- I think in the conclusion we explicitly state that software teams should make decisions about whether or not to adopt these different tools/approaches rather than be told by us. For instance, I have worked for a long time with algorithmic approaches that are deeply beneficial and were created long before current chat bots (eg, automated error detection in hardware processes)
@Di4na that being said, given that many people's teams ARE using tools like copilot I think it is valid inquiry to consider it a form of workplace upskilling -- as you see we also do a lot of work in this paper to show that many developers are questioning the quality of output from these tools and that's one major recommendation to orgs, to include that perspective. What I consider upskilling might result in a team decision to NOT adopt a tool. eg, upskilling in security is an explicit example
@Di4na but I recognize that these are highly difficult topics with a broad range of possible interpretations, and I know that scientific writing can sometimes have too much detachment. Since this is a working paper I have taken this feedback and will be thinking about how to add clarity on the interpretation! Thank you so much for sharing.
@Di4na on a personal note as a queer person working on this, I feel deeply that many large-scale applications of what we call AI are damaging, but as a psychologist I also wanted to provide support for teams surrounded by many types of tools, and I don't believe every application of automation and prediction in our work is wrong -- we think that what we've provided support for in this paper is good psych culture to help teams feel self-efficacy in diagnosing coding tools specifically
@grimalkina I think deep down, it is the whole talk of "the transition" as if inevitable that feels like this. Our reading around my networks is that there is... Nothing to transition into and that these tools don't impact a lot.
Which makes reading this feel like I am on another planet.
Which tbf. May just be because of the sheer size of the industry.
@Di4na objectively speaking it is true that large numbers of developers are using tools like copilot (eg recent stack overflow survey) altho how & when, much to be learned! In my friends' coding courses and education work it is also pervasive that students are trying to figure out whether and how to adopt these tools and some are very scared of being "left behind" by workplaces making this type of thing a mandate. That conflict between perceptions and expectations and reality is psychology :)
@Di4na I also feel it's kind of interesting to look at this like a question about "how software teams are impacted by a lot of hype and pressure to change" whether or not we actually think a change is real or going to happen vs going to disintegrate -- because this could apply for many technological "moments"!
I love this study. Rather than focus on the tool, we're focusing on the humans and what they are experiencing while thinking about other people thinking about their skills. Rather than presuming that we as researchers get to tell software teams what to do, we hope this work helps software teams work through the process of healthier decision-making so THEY can figure out what they want to do. We have also worked very hard to be able to observe critical equity and opportunity gaps.
In accordance with our values as action researchers, and our lives as applied researchers (not academics) trying to share work to a public audience in a timely manner, we choose to do this work and share it with you rapidly. However, I would really love to submit this paper to a journal. I am really proud of this work and the impact it's already had with real teams. I have no doubt it will be a struggle finding a software venue that can review these methods and will publish equity gap findings.
So, you know, let me know if you want to invite this somewhere. Work that's been presented to multiple engineering leadership teams already and changed policies in multiple technology orgs should be relevant to someone :)
Add comment