User Research in a “New Age of Computing”


alexa personal digital assitant

This is my submission for the challenge report submission in the course user research methods. It tries to answer the question about the use and role of user research in the “new age of computing we are entering into”.

The “New Age of Computing”

What about user research and the new age of computing that we are entering into? Is there still a role for user research?

I think a large part about the “uncertainty” of user researches role in this purported new age of computing is related to the advances in machine learning and artificial intelligence and a vision that these could automate the gathering of data and design/generating of interfaces. That or from a (mistaken assumption) that some interfaces don't require design any more because they're e.g. using natural language interfaces and everybody should be able to interact with any of these by default.

“Not needing design” in natural language interfaces is relatively easily refuted: The current variants still struggle with contextual information and the inherent unpreciseness of human language, even for so trivial things as “remind me later this week” (Rong et al. 2017)1. Also, in their current state, they are – in one aspect – a step back to command line interfaces: they require knowledge which commands are available and what can be expected from the system. Lastly, their notorious problems with accents have become quite famous (Scottish Elevator With Voice Recognition 2017)2. All of these mean that research on the context is required and design decisions made, that need to be informed.

Another way of “not needing design” would be the scenario where a given task is taken out of the users hands entirely, as is the vision for autonomous cars. Even if the vision came to pass, there'd still be interfaces – e.g. specifying where the car should go – as well as system properties to design, that require knowledge about the people and their context and thus user research. This should hold true, as long as there are humans and technology as there always will be a point where they interact. However, in any way, the current reality of (semi-)autonomous cars is far from that, as these cars require constant oversight and thoughtful ways of signaling and dealing with handover (van der Heiden, Iqbal, and Janssen 2017)3. If these aren't handled well, problems in the engine can easily lead to dangerous situations (Brown and Laurier 2017)4.

The more probable scenario, but I'd argue still unlikely or at least very far off future, is that research and design will be fully automatized with the use of big data, modern machine learning and extensive personalisation, adaptive interfaces-for-one as well as procedural content generation and computational creativity. However, in all of these cases the algorithms need massive tweaking of meta-parameters, a careful selection of a training set, as well as an integration with other interface methods (think of e.g. filtering and sorting of Facebook-feed with the rest of Facebook-app and the post-primitives). So in most cases, I'd argue these technologies are better seen as tools during the research phase (e.g. most things that fall under data science), as tools to build interfaces with (e.g. recommender systems, procedural generation) or as features of the artifact itself (e.g. online style-transfer on images^[e.g. https://deepart.io/] that use deep neural networks in particular auto-encoders in the background)

So, I'd argue that these technologies, just like the many others that are currently developed (e.g. affordable VR/AR-HMDs) or already exist, are either tools that aid research, aid design or are used as part of the designed artifact or system itself.

Thus, I think the more interesting question about this “New Age of Computing” is not if user research will still be a thing, but what thing it – and HCI-research in general – should be in it, what role it should play. This applies particularly in regard to the larger societal issues arising around the use of designed technologies, e.g. the vanishing of low-skill jobs due to automation, the digital divide, power-dynamics and terrible work reality in crowd-work, filterbubbles and their effect on democracy, algorithmic bias and accountability, (mass) surveillance and manipulative design5. In this regard, Ben Schneiderman (2017) lists the following “Grand Challenges” for HCI (and thus user research as integral part of it):

  • Develop a handbook of human needs
  • Shift from user experience to community experience
  • Refine theories of persuasion
  • Encourage resource conservation
  • Shape the learning health system
  • Advance the design of medical devices
  • Support successful aging strategies
  • Promote lifelong learning
  • Stimulate rapid interface learning [e.g. via multilayer interfaces]
  • Engineer new business models
  • Design novel input and output devices
  • Accelerate analytic clarity [by utilizing big data]
  • Amplify empathy, compassion, and caring
  • Secure cyberspace
  • Encourage reflection, calmness, and mindfulness
  • Clarify responsibility and accountability [especially with autonomous machines]

The Role of User Researchers

What is this role and how will it change? What could it contribute, what ways would it need to be adapted?

Brownlee (2016)6 argues that a few roles will fade out or lose in significance, will be split up or merged into others, as well as some others will emerge: He argues that UX Designer as a job description will be applied less often, as it's become a very broad umbrella term anyway. Similarly, he thinks that Design Research has become so fundamental for everyone, that they demand for people only doing that might recede. Furthermore, in his opinion, visual designers will (more) heavily rely on algorithms and procedural generation, (Post-)Industrial designers will (need to) increasingly design computerized products. As for new design- and research-jobs he points out: the virtual interaction designer, the specialist material designer (e.g. 3d printing and smart materials), the algorithm- and AI-design specialist, the design strategist that informs policy-makers, organization designers, and most and foremost all of the above as freelancers.

John Brownlee isn't the only one pointing out the increasing integration of novel algorithms with the design and research processes:

Dove et al. (2017)7 talks about the design problems and challenges of machine-learning systems, studies how the developer- and design-teams interact and especially points out a lack/insufficiency of machine learning frameworks suited for rapid prototyping by designers, the lack of material that explain the possibilities and limits of ML to them and the unrefinedness from a design-perspective of the “medium” (as can be seen e.g. in the many problems digital personal assistants and speech interfaces in general still struggle with).

I think extensive and rigorous user research (as part of design iterations) might help somewhat with that unrefinedness. In general, it could do this via study of which algorithms and parameters work for which use cases, what their behaviours are, etc. In particular, on a project level, user research can inform what good and bad example generated artifacts are from which rules can be derived or against which algorithm-parameters can be fine-tuned, i.e. the critical step in designing generators (Kompton 2017)8 and I'd argue machine learning systems as well.

Thus for people doing user research, the new technologies could mean additional tools for their belt as well additional subject matter to research around.

As with the “new age” section above, I also think there's a more interesting question here as well: instead of asking for the things that “will” be, i.e. that can be said to happen with any degree of certainty, there's the question of what their role (and tackled challenges) could be, or even more so, should be.

As far as the challenges are concerened, I'd like to point to the (societal) issues at the end of the section above and by Shneiderman et al. (2017)9. In regard to these I'd like to stress (user) researchers’ role in informing policy-makers to make responsible decisions on technology regulation, use and development. On this Pargman et al Pargman et al. (2017)10 write

It should however be remembered that it is not uncommon in decision making to only refer to “what will happen” (for example “you can’t stop innovation so you shouldn't even try”), as if it was impossible to decide what future we want and try to influence events by working towards the realization of particular futures.

Pargman et al call these “will”-questions predictive scenarios and argue for the use of explorative scenarios (“what can happen?") and normative scenarios (“what should happen?") in the research informing policy-decisions.

In regard to changing research methodology, some of Shneiderman's Grand Challenges latter also argue for that and a change in the underlying theory of doing user research itself, with an suggested increased focus on instrumentalizing big data, as well as theories of human needs and persuasiveness, and empathy, compassion, caring as well as community experience. A trend that is already occuring is dropping the term “user” in favor of a more general “human” to break up this power-structure and broaden the horizon accordingly. Baumer and Brubaker (2017)11 pick this up by analyzing the “user”-concept throughout the three waves and arguing for post-userism, i.e. looking beyond the users represented in system/data-structures, user-interfaces, design-processes as well as design- and research-ideology. They point out five scenarios that the classical user-concept wouldn't capture:

  • indirection: people interacting with system via others
  • transience: no distinction between multiple interactions by one or many persons, e.g. public displays
  • multiplicity: one person with multiple user-accounts for multiple roles
  • absence: e.g. people without an fb-account are still impacted by it
  • hybrid: some actors can be seen as in-between of organization, AI and human individual. for a very crude example: semi-automated customer service

Conclusion

Concluding, I'd argue that, yes, user research will still play a role, as long as there are humans interacting with and being influenced by technology. The developments that constitute this “New Age of Computing” should be seen as providing new tools for research, design, development and as basis of build artifact, but also are technologies and method from which new societal challenges arise, that need to be studied and addressed.


  1. Rong, Xin, Adam Fourney, Robin N. Brewer, Meredith Ringel Morris, and Paul N. Bennett. 2017. “Managing Uncertainty in Time Expressions for Virtual Assistants.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 568–79. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025674. ↩︎

  2. Scottish Elevator With Voice Recognition. 2017. Burnsitown Comedy Show. Accessed December 29. https://www.youtube.com/watch?v=BOUTfUmI8vs. ↩︎

  3. van der Heiden, Remo M.A., Shamsi T. Iqbal, and Christian P. Janssen. 2017. “Priming Drivers Before Handover in Semi-Autonomous Cars.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 392–404. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025507. ↩︎

  4. Brown, Barry, and Eric Laurier. 2017. “The Trouble with Autopilots: Assisted and Autonomous Driving on the Social Road.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 416–29. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025462. ↩︎

  5. for a very extreme case see Sesame Credit, a system heavily relying on gamification and operand conditioning, where friends with bad ratings pull you down with them, and where those ratings could soon determine access to jobs, social security, visa, etc (Nguyen 201712) ↩︎

  6. Brownlee, John. 2016. “5 Design Jobs That Won’t Exist In The Future.” Co.Design. September 1. https://www.fastcodesign.com/3063318/5-design-jobs-that-wont-exist-in-the-future. ↩︎

  7. Dove, Graham, Kim Halskov, Jodi Forlizzi, and John Zimmerman. 2017. “UX Design Innovation: Challenges for Working with Machine Learning As a Design Material.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 278–88. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025739. ↩︎

  8. Kompton, Kate. 2017. “So You Want to Build a Generator….” Kate Compton. Accessed December 29. http://galaxykate0.tumblr.com/post/139774965871/so-you-want-to-build-a-generator. ↩︎

  9. Shneiderman, Ben, Catherine Plaisant, Maxine Cohen, Steven Jacobs, Niklas Elmqvist, and Nicholoas Diakopoulos. 2017. “Grand Challenges for HCI Researchers | ACM Interactions.” Accessed December 22. http://interactions.acm.org/archive/view/september-october-2016/grand-challenges-for-hci-researchers. ↩︎

  10. Pargman, Daniel, Elina Eriksson, Mattias Höjer, Ulrika Gunnarsson Östling, and Luciane Aguiar Borges. 2017. “The (Un)Sustainability of Imagined Future Information Societies.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 773–85. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025858. ↩︎

  11. Baumer, Eric P. S., and Jed R. Brubaker. 2017. “Post-Userism.” In Proceedings of the 2017 CHI Conference on Human Factors in Computing Systems, 6291–6303. CHI ’17. New York, NY, USA: ACM. doi:10.1145 / 3025453.3025740. ↩︎

  12. Nguyen, Clinton. 2017. “China Might Use Data to Create a Score for Each Citizen Based on How Trustworthy They Are.” Business Insider Deutschland. Accessed December 29. http://www.businessinsider.de/china-social-credit-score-like-black-mirror-2016-10. ↩︎