Participatory AI literature review

I’ve been think­ing alot about civic par­tic­i­pa­tion in machine learn­ing sys­tems devel­op­ment. In par­tic­u­lar, involv­ing non-experts in the poten­tial­ly val­ue-laden trans­la­tion work from spec­i­fi­ca­tions that engi­neers do when they build their mod­els. Below is a sum­ma­ry of a selec­tion of lit­er­a­ture I found on the top­ic, which may serve as a jump­ing-off point for future research.

Abstract

The lit­er­a­ture on par­tic­i­pa­to­ry arti­fi­cial intel­li­gence (AI) reveals a com­plex land­scape marked by chal­lenges and evolv­ing method­olo­gies. Fef­fer et al. (2023) cri­tique the reduc­tion of par­tic­i­pa­tion to com­pu­ta­tion­al mech­a­nisms that only approx­i­mate nar­row moral val­ues. They also note that engage­ments with stake­hold­ers are often super­fi­cial and unrep­re­sen­ta­tive. Groves et al. (2023) iden­ti­fy sig­nif­i­cant bar­ri­ers in com­mer­cial AI labs, includ­ing high costs, frag­ment­ed approach­es, exploita­tion con­cerns, lack of trans­paren­cy, and con­tex­tu­al com­plex­i­ties. These bar­ri­ers lead to a piece­meal approach to par­tic­i­pa­tion with min­i­mal impact on deci­sion-mak­ing in AI labs. Del­ga­do et al. (2023) observe that par­tic­i­pa­to­ry AI involves stake­hold­ers most­ly in a con­sul­ta­tive role with­out inte­grat­ing them as active deci­sion-mak­ers through­out the AI design lifecycle.

Gerdes (2022) pro­pos­es a data-cen­tric approach to AI ethics and under­scores the need for inter­dis­ci­pli­nary bridge builders to rec­on­cile dif­fer­ent stake­hold­er per­spec­tives. Robert­son et al. (2023) explore par­tic­i­pa­to­ry algo­rithm design, empha­siz­ing the need for pref­er­ence lan­guages that bal­ance expres­sive­ness, cost, and collectivism—Sloane et al. (2020) cau­tion against “par­tic­i­pa­tion wash­ing” and the poten­tial for exploita­tive com­mu­ni­ty involve­ment. Brat­teteig & Verne (2018) high­light AI’s chal­lenges to tra­di­tion­al par­tic­i­pa­to­ry design (PD) meth­ods, includ­ing unpre­dictable tech­no­log­i­cal changes and a lack of user-ori­ent­ed eval­u­a­tion. Birhane et al. (2022) call for a clear­er under­stand­ing of mean­ing­ful par­tic­i­pa­tion, advo­cat­ing for a shift towards vibrant, con­tin­u­ous engage­ment that enhances com­mu­ni­ty knowl­edge and empow­er­ment. The lit­er­a­ture sug­gests a press­ing need for more effec­tive, inclu­sive, and empow­er­ing par­tic­i­pa­to­ry approach­es in AI development.

Bibliography

  1. Birhane, A., Isaac, W., Prab­hakaran, V., Diaz, M., Elish, M. C., Gabriel, I., & Mohamed, S. (2022). Pow­er to the Peo­ple? Oppor­tu­ni­ties and Chal­lenges for Par­tic­i­pa­to­ry AI. Equi­ty and Access in Algo­rithms, Mech­a­nisms, and Opti­miza­tion, 1–8. https://doi.org/10/grnj99
  2. Brat­teteig, T., & Verne, G. (2018). Does AI make PD obso­lete?: Explor­ing chal­lenges from arti­fi­cial intel­li­gence to par­tic­i­pa­to­ry design. Pro­ceed­ings of the 15th Par­tic­i­pa­to­ry Design Con­fer­ence: Short Papers, Sit­u­at­ed Actions, Work­shops and Tuto­r­i­al — Vol­ume 2, 1–5. https://doi.org/10/ghsn84
  3. Del­ga­do, F., Yang, S., Madaio, M., & Yang, Q. (2023). The Par­tic­i­pa­to­ry Turn in AI Design: The­o­ret­i­cal Foun­da­tions and the Cur­rent State of Prac­tice. Pro­ceed­ings of the 3rd ACM Con­fer­ence on Equi­ty and Access in Algo­rithms, Mech­a­nisms, and Opti­miza­tion, 1–23. https://doi.org/10/gs8kvm
  4. Ehsan, U., & Riedl, M. O. (2020). Human-Cen­tered Explain­able AI: Towards a Reflec­tive Sociotech­ni­cal Approach. In C. Stephani­dis, M. Kuro­su, H. Degen, & L. Rein­er­man-Jones (Eds.), HCI Inter­na­tion­al 2020—Late Break­ing Papers: Mul­ti­modal­i­ty and Intel­li­gence (pp. 449–466). Springer Inter­na­tion­al Pub­lish­ing. https://doi.org/10/gskmgf
  5. Fef­fer, M., Skir­pan, M., Lip­ton, Z., & Hei­dari, H. (2023). From Pref­er­ence Elic­i­ta­tion to Par­tic­i­pa­to­ry ML: A Crit­i­cal Sur­vey & Guide­lines for Future Research. Pro­ceed­ings of the 2023 AAAI/ACM Con­fer­ence on AI, Ethics, and Soci­ety, 38–48. https://doi.org/10/gs8kvx
  6. Gerdes, A. (2022). A par­tic­i­pa­to­ry data-cen­tric approach to AI Ethics by Design. Applied Arti­fi­cial Intel­li­gence, 36(1), 2009222. https://doi.org/10/gs8kt4
  7. Groves, L., Pep­pin, A., Strait, A., & Bren­nan, J. (2023). Going pub­lic: The role of pub­lic par­tic­i­pa­tion approach­es in com­mer­cial AI labs. Pro­ceed­ings of the 2023 ACM Con­fer­ence on Fair­ness, Account­abil­i­ty, and Trans­paren­cy, 1162–1173. https://doi.org/10/gs8kvs
  8. Robert­son, S., Nguyen, T., Hu, C., Albis­ton, C., Nikzad, A., & Sale­hi, N. (2023). Expres­sive­ness, Cost, and Col­lec­tivism: How the Design of Pref­er­ence Lan­guages Shapes Par­tic­i­pa­tion in Algo­rith­mic Deci­sion-Mak­ing. Pro­ceed­ings of the 2023 CHI Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems, 1–16. https://doi.org/10/gr6q2t
  9. Sloane, M., Moss, E., Awom­o­lo, O., & For­lano, L. (2020). Par­tic­i­pa­tion is not a Design Fix for Machine Learn­ing. arXiv:2007.02423 [Cs]. http://arxiv.org/abs/2007.02423
  10. Zytko, D., J. Wis­niews­ki, P., Guha, S., P. S. Baumer, E., & Lee, M. K. (2022). Par­tic­i­pa­to­ry Design of AI Sys­tems: Oppor­tu­ni­ties and Chal­lenges Across Diverse Users, Rela­tion­ships, and Appli­ca­tion Domains. Extend­ed Abstracts of the 2022 CHI Con­fer­ence on Human Fac­tors in Com­put­ing Sys­tems, 1–4. https://doi.org/10/gs8kv6

Playful Design for Workplace Change Management’ at PLAYTrack conference 2017 in Aarhus

Lase defender collab at FUSE

At the end of last year I was invit­ed to speak at the PLAY­Track con­fer­ence in Aarhus about the work­place change man­age­ment games made by Hub­bub. It turned out to be a great oppor­tu­ni­ty to recon­nect with the play research community. 

I was very much impressed by the pro­gram assem­bled by the organ­is­ers. Peo­ple came from a wide range of dis­ci­plines and cru­cial­ly, there was ample time to dis­cuss and reflect on the mate­ri­als pre­sent­ed. As I tweet­ed after­wards, this is a thing that most con­fer­ence organ­is­ers get wrong.

I was par­tic­u­lar­ly inspired by the work of Ben­jamin Mardell and Mara Krechevsky at Harvard’s Project ZeroMak­ing Learn­ing Vis­i­ble looks like a great resource for any­one who teach­es. Then there was Reed Stevens from North­west­ern Uni­ver­si­ty whose project FUSE is one of the most sol­id exam­ples of play­ful learn­ing for STEAM I’ve seen thus far. I was also fas­ci­nat­ed by Cia­ra Laverty’s work at PEDAL on observ­ing par­ent-child play. Miguel Sicart deliv­ered anoth­er great provo­ca­tion on the dark side of play­ful design. And final­ly I was delight­ed to hear about and expe­ri­ence for myself some of Amos Blan­ton’s work at the LEGO Foun­da­tion. I should also call out Ben Fin­cham’s many provoca­tive con­tri­bu­tions from the audience.

The abstract for my talk is below, which cov­ers most of what I talked about. I tried to give peo­ple a good sense of: 

  • what the games con­sist­ed of,
  • what we were aim­ing to achieve,
  • how both the fic­tion and the play­er activ­i­ties sup­port­ed these goals,
  • how we made learn­ing out­comes vis­i­ble to our play­ers and clients,
  • and final­ly how we went about design­ing and devel­op­ing these games.

Both projects have sol­id write-ups over at the Hub­bub web­site, so I’ll just point to those here: Code 4 and Rip­ple Effect.

In the final sec­tion of the talk I spent a bit of time reflect­ing on how I would approach projects like this today. After all, it has been sev­en years since we made Code 4, and four years since Rip­ple Effect. That’s ages ago and my per­spec­tive has def­i­nite­ly changes since we made these.

Participatory design

First of all, I would get even more seri­ous about co-design­ing with play­ers at every step. I would recruit rep­re­sen­ta­tives of play­ers and invest them with real influ­ence. In the projects we did, the pri­ma­ry vehi­cle for play­er influ­ence was through playtest­ing. But this is nec­es­sar­i­ly lim­it­ed. I also won’t pre­tend this is at all easy to do in a com­mer­cial context. 

But, these games are ulti­mate­ly about improv­ing work­er pro­duc­tiv­i­ty. So how do we make it so that work­ers share in the real-world prof­its yield­ed by a suc­cess­ful cul­ture change?

I know of the exis­tence of par­tic­i­pa­to­ry design but from my expe­ri­ence it is not a com­mon approach in the indus­try. Why?

Value sensitive design

On a relat­ed note, I would get more seri­ous about what val­ues are sup­port­ed by the sys­tem, in whose inter­est they are and where they come from. Ear­ly field research and work­shops with audi­ence do sur­face some val­ues but val­ues from cus­tomer rep­re­sen­ta­tives tend to dom­i­nate. Again, the com­mer­cial con­text we work in is a poten­tial challenge. 

I know of val­ue sen­si­tive design, but as with par­tic­i­pa­to­ry design, it has yet to catch on in a big way in the indus­try. So again, why is that?

Disintermediation

One thing I con­tin­ue to be inter­est­ed in is to reduce the com­plex­i­ty of a game system’s phys­i­cal affor­dances (which includes its code), and to push even more of the sub­stance of the game into those social allowances that make up the non-mate­r­i­al aspects of the game. This allows for spon­ta­neous rene­go­ti­a­tion of the game by the play­ers. This is dis­in­ter­me­di­a­tion as a strat­e­gy. David Kanaga’s take on games as toys remains huge­ly inspi­ra­tional in this regard, as does Bernard De Koven’s book The Well Played Game.

Gamefulness versus playfulness

Code 4 had more focus on sat­is­fy­ing the need for auton­o­my. Rip­ple Effect had more focus on com­pe­tence, or in any case, it had less empha­sis on auton­o­my. There was less room for ‘play’ around the core dig­i­tal game. It seems to me that mas­ter­ing a sub­jec­tive sim­u­la­tion of a sub­ject is not nec­es­sar­i­ly what a work­place game for cul­ture change should be aim­ing for. So, less game­ful design, more play­ful design.

Adaptation

Final­ly, the agency mod­el does not enable us to stick around for the long haul. But work­place games might be bet­ter suit­ed to a set­up where things aren’t thought of as a one-off project but more of an ongo­ing process. 

In How Build­ings Learn, Stew­art Brand talks about how archi­tects should revis­it build­ings they’ve designed after they are built to learn about how peo­ple are actu­al­ly using them. He also talks about how good build­ings are build­ings that its inhab­i­tants can adapt to their needs. What does that look like in the con­text of a game for work­place cul­ture change?


Play­ful Design for Work­place Change Management

Code 4 (2011, com­mis­sioned by the Tax Admin­is­tra­tion of the Nether­lands) and Rip­ple Effect (2013, com­mis­sioned by Roy­al Dutch Shell) are both games for work­place change man­age­ment designed and devel­oped by Hub­bub, a bou­tique play­ful design agency which oper­at­ed from Utrecht, The Nether­lands and Berlin, Ger­many between 2009 and 2015. These games are exam­ples of how a goal-ori­ent­ed seri­ous game can be used to encour­age play­ful appro­pri­a­tion of work­place infra­struc­ture and social norms, result­ing in an open-end­ed and cre­ative explo­ration of new and inno­v­a­tive ways of working.

Seri­ous game projects are usu­al­ly com­mis­sioned to solve prob­lems. Solv­ing the prob­lem of cul­tur­al change in a straight­for­ward man­ner means view­ing games as a way to per­suade work­ers of a desired future state. They typ­i­cal­ly take videogame form, sim­u­lat­ing the desired new way of work­ing as deter­mined by man­age­ment. To play the game well, play­ers need to mas­ter its sys­tem and by extension—it is assumed—learning happens.

These games can be be enjoy­able expe­ri­ences and an improve­ment on pre­vi­ous forms of work­place learn­ing, but in our view they decrease the pos­si­bil­i­ty space of poten­tial work­place cul­tur­al change. They dimin­ish work­er agency, and they waste the cre­ative and inno­v­a­tive poten­tial of involv­ing them in the inven­tion of an improved work­place culture. 

We instead choose to view work­place games as an oppor­tu­ni­ty to increase the space of pos­si­bil­i­ty. We resist the temp­ta­tion to bake the desired new way of work­ing into the game’s phys­i­cal and dig­i­tal affor­dances. Instead, we leave how to play well up to the play­ers. Since these games are team-based and col­lab­o­ra­tive, play­ers need to nego­ti­ate their way of work­ing around the game among them­selves. In addi­tion, because the games are dis­trib­uted in time—running over a num­ber of weeks—and are playable at play­er dis­cre­tion dur­ing the work­day, play­ers are giv­en license to appro­pri­ate work­place infra­struc­ture and sub­vert social norms towards in-game ends.

We tried to make learn­ing tan­gi­ble in var­i­ous ways. Because the games at the core are web appli­ca­tions to which play­ers log on with indi­vid­ual accounts we were able to col­lect data on play­er behav­iour. To guar­an­tee pri­va­cy, employ­ers did not have direct access to game data­bas­es and only received anonymised reports. We took respon­si­bil­i­ty for play­er learn­ing by facil­i­tat­ing coach­ing ses­sions in which they could safe­ly reflect on their game expe­ri­ences. Round­ing out these efforts, we con­duct­ed sur­veys to gain insight into the play­er expe­ri­ence from a more qual­i­ta­tive and sub­jec­tive perspective.

These games offer a mod­el for a rea­son­ably demo­c­ra­t­ic and eth­i­cal way of doing game-based work­place change man­age­ment. How­ev­er, we would like to see efforts that fur­ther democ­ra­tise their design and development—involving work­ers at every step. We also wor­ry about how games can be used to cre­ate the illu­sion of work­er influ­ence while at the same time soft­ware is deployed through­out the work­place to lim­it their agency. 

Our exam­ples may be inspir­ing but because of these devel­op­ments we feel we can’t con­tin­ue this type of work with­out seri­ous­ly recon­sid­er­ing our cur­rent process­es, tech­nol­o­gy stacks and busi­ness practices—and ulti­mate­ly whether we should be mak­ing games at all.

Game player needs and designing architectures of participation

How do you cre­ate a cor­po­rate envi­ron­ment in which peo­ple share knowl­edge out of free will?1 This is a ques­tion my good friends of Wemind2 are work­ing to answer for their clients on a dai­ly basis.3 We’ve recent­ly decid­ed to col­lab­o­ra­tive­ly devel­op meth­ods use­ful for the design of a par­tic­i­pa­to­ry con­text in the work­place. Our idea is that since knowl­edge shar­ing is essen­tial­ly about peo­ple inter­act­ing in a con­text, we’ll apply inter­ac­tion design meth­ods to the prob­lem. Of course, some meth­ods will be more suit­ed to the prob­lem than oth­ers, and all will need to be made spe­cif­ic for them to real­ly work. That’s the challenge.

Nat­u­ral­ly I will be look­ing for inspi­ra­tion in game design the­o­ry. This gives me a good rea­son to blog about the PENS mod­el. I read about this in an excel­lent Gama­su­tra arti­cle titled Rethink­ing Car­rots: A New Method For Mea­sur­ing What Play­ers Find Most Reward­ing and Moti­vat­ing About Your Game. The cre­ators of this mod­el4 want­ed to bet­ter under­stand what fun­da­men­tal­ly moti­vates game play­ers as well as come up with a prac­ti­cal play test­ing mod­el. What they’ve come up with is intrigu­ing: They’ve demon­strat­ed that to offer a fun expe­ri­ence, a game has to sat­is­fy cer­tain basic human psy­cho­log­i­cal needs: com­pe­tence, auton­o­my and relat­ed­ness.5

I urge any­one inter­est­ed in what makes games work their mag­ic to read this arti­cle. It’s real­ly enlight­en­ing. The cool thing about this mod­el is that it pro­vides a deep­er vocab­u­lary for talk­ing about games.6 In the arti­cle’s con­clu­sion the authors note the same, and point out that by using this vocab­u­lary we can move beyond cre­at­ing games that are ‘mere’ enter­tain­ment. They men­tion seri­ous games as an obvi­ous area of appli­ca­tion, I can think of many more (3C prod­ucts for instance). But I plan on apply­ing this under­stand­ing of game play­er needs to the design of archi­tec­tures of par­tic­i­pa­tion. Wish me luck.

  1. Tra­di­tion­al­ly, shar­ing knowl­edge in large organ­i­sa­tions is explic­it­ly reward­ed in some way. Arguably true knowl­edge can only be shared vol­un­tar­i­ly. []
  2. Who have been so kind to offer me some free office space, Wi-Fi and cof­fee since my arrival in Copen­hagen. []
  3. They are par­tic­u­lar­ly focused on the val­ue of social soft­ware in this equa­tion. []
  4. Scott Rig­by and Richard Ryan of Immer­syve []
  5. To nuance this, the amount to which a play­er expects each need to be sat­is­fied varies from game genre to genre. []
  6. Sim­i­lar to the work of Koster and of Salen & Zim­mer­man. []