FRUGAL AI FOR INCLUSIVE FUTURES: ENABLING GRASSROOTS INNOVATION IN HEALTH, EDUCATION, AND GOVERNANCE

59 CHAPTER-6 FRUGAL AI FOR INCLUSIVE FUTURES: ENABLING GRASSROOTS INNOVATION IN HEALTH, EDUCATION, AND GOVERNANCE D . M Zahee Ahmed Assis an P o esso and Head PG Depa men o Comme ce and Resea ch D Kalaigna M. Ka unanidhi Go e nmen Ins i u e o Pos G adua e S udies and Resea ch, Ka aikal, U.T o Puduche y, India Abs ac A i icial In elligence (AI) can only be uly ans o ma i e when i is accessible, e hical, and oo ed in he li ed eali ies o di e se communi ies. This chap e explo es he concep o F ugalAI a low-cos , esou ce-sensi i e, and inclusi e app oach o digi al inno a ion and i s applica ion ac oss sec o s such as heal hca e, educa ion, and go e nance. The s udy ocuses on he Union Te i o y o Puduche y as a pilo si e, le e aging i s unique Indo–F ench legacy, mul ilingual popula ion, and policy agili y o design and es scalable AI solu ions. By p oposing he Indo–F ench F ugal AI Inno a ion Hub (IFFAIH), he chap e demons a es how global pa ne ships, diaspo a engagemen , and g ass oo s pa icipa ion can con e ge o co-c ea e AI ools o ci ic se ices, mul ilingual educa ion, eleheal h, and g ie ance ed essal. Aligned wi h Sus ainable De elopmen Goals (SDGs), he chap e maps exis ing Cen al and S a e go e nmen schemes o suppo implemen a ion, ensu ing cos -e ec i eness and cul u al ele ance. Th ough policy ecommenda ions, a localized oadmap, and an ins i u ional model, his chap e posi ions Puduche y as a eplicable es bed o e hical, inclusi e, and human-cen ic AI inno a ion demons a ing ha he u u e o AI mus no only be sma , bu also socially jus and locally g ounded. 1. In oduc ion In he e a o apid digi al ans o ma ion, A i icial In elligence (AI) has eme ged as bo h a dis up o and an enable ac oss domains such as heal hca e, educa ion, go e nance, and li elihoods. While much o AI discou se is cen e ed a ound high- ech inno a ion and esou ce-in ensi e models, he p essing need in he Global Sou h lies in de eloping con ex ual, low-cos , and scalable AI solu ions ha p io i ize accessibili y and social equi y. This calls o a pa adigm shi owa d wha is inc easingly known as F ugal AI a model ha combines echnological e iciency wi h economic and cul u al ele ance. This chap e c i ically examines how F ugal AI can se e as a s a egic enable o inclusi e de elopmen . I ad oca es o e hical, localized, and mul ilingual AI models ha empowe unde se ed popula ions, pa icula ly in egions ma ked by adminis a i e uniqueness, linguis ic di e si y, and his o ical linkages o global 60 sys ems. In doing so, i highligh s he Union Te i o y o Puduche y as a li ing labo a o y o es ing AI-d i en public se ice ools gi en i s bilingual he i age, dynamic diaspo a, and p oximi y o Indo–F ench ins i u ional ecosys ems. G ounded in he Sus ainable De elopmen Goals (SDGs), he chap e aims o p opose a eplicable inno a ion amewo k ha links ugal echnological design wi h g ass oo s applica ion, ci ic engagemen , and ansna ional coope a ion. The goal is no me ely o ad ance AI pene a ion, bu o eimagine i as a people- cen ic, c oss-cul u al, and economically iable o ce o sus ainable ans o ma ion. 2. Concep ual F amewo k and Li e a u e Re iew 2.1 F ugal AI: A Pa adigm o Inclusi e Inno a ion F ugal AI is an eme ging design philosophy ha emphasizes c ea ing esou ce- e icien , con ex -sensi i e, and a o dable AI solu ions o unde se ed communi ies. Unlike con en ional AI models de eloped in da a- ich and capi al- in ensi e en i onmen s, F ugal AI p io i izes simplici y, cos -e ec i eness, and adap abili y. Schola s such as Radjou and P abhu (2012) ha e a gued ha ugal inno a ion, o en bo n ou o sca ci y, can se e as a sp ingboa d o scalable ans o ma ion in de eloping na ions. Applied o AI, his means de eloping ools ha equi e minimal compu a ional esou ces, ely on open da ase s, and can ope a e ac oss low-connec i i y egions. The concep aligns closely wi h digi al public goods, low- esou ce AI models, and human-cen e ed compu ing, and is inc easingly ecognized as essen ial o achie ing he SDGs, pa icula ly hose ela ed o educa ion (SDG 4), heal hca e (SDG 3), gende equali y (SDG 5), and indus y inno a ion (SDG 9). I also esona es wi h UNESCO’s call o AI ha upholds human igh s, di e si y, and equi y. 2.2 Socio-Technological Con ex in India and Global Sou h India, wi h i s as linguis ic di e si y, u al–u ban di ide, and dynamic go e nance models, p esen s bo h a challenge and an oppo uni y o AI localiza ion. Na ional p og ams such as he Digi al India Mission, Ayushman Bha a Digi al Heal h Mission, and PM Ga i Shak i ha e c ea ed enabling amewo ks o deploying AI in heal hca e, logis ics, ag icul u e, and ci izen se ices. Howe e , e ec i e deploymen is o en hinde ed by in as uc u e gaps, digi al illi e acy, and linguis ic ba ie s unde sco ing he need o F ugal AI ailo ed o local con ex s. Globally, coun ies in he Global Sou h a e expe imen ing wi h AI-on- he-edge, communi y-anno a ed da ase s, and open-sou ce machine lea ning pla o ms o add ess egion-speci ic needs. Ini ia i es like Sma A ica, Aap i Ins i u e’s AI Localism epo s, and GSMA’s Mobile o De elopmen p og am o e aluable insigh in o he iabili y o cos -sensi i e AI. 61 2.3 Cul u al Con ex and Indo–F ench Oppo uni ies Puduche y’s F anco–Indian he i age adds a unique cul u al dimension o AI de elopmen . Wi h i s ins i u ional connec ions o F ance, mul ilingualism (Tamil, F ench, and English), and access o diaspo a ne wo ks, Puduche y becomes an ideal g ound o ansna ional AI pilo p ojec s. The po en ial o co- de elop ools in collabo a ion wi h F ench uni e si ies, he Alliance F ançaise, and diaspo a echnologis s adds global c edibili y o he o he wise local expe imen a ion se ing he ounda ion o he Indo–F ench F ugal AI Inno a ion Hub p oposed in his chap e . 2.4 Re iew o F ugal AI in P ac ice: India and he Global Sou h Recen yea s ha e wi nessed a su ge in ugal echnological in e en ions, pa icula ly in egions whe e adi ional in as uc u e and access a e limi ed. In India, ini ia i es like eSanjee ani (a elemedicine pla o m) and DIKSHA (a mul ilingual digi al lea ning eposi o y) exempli y how low-cos digi al solu ions can b idge gaps in public se ice deli e y. These pla o ms ope a e on open s anda ds, equi e minimal bandwid h, and u ilize de ices al eady a ailable o use s p inciples ha align closely wi h F ugal AI design. Beyond India, se e al coun ies in he Global Sou h ha e emb aced simila models. Fo example: Sma A ica has p omo ed AI-d i en ma e nal ca e ale s h ough SMS bo s in Uganda and Rwanda. In Bangladesh, AI models ained on local loodplain da a ha e helped in disas e p epa edness wi h minimal compu a ional cos . Kenya’s AI-powe ed soil diagnos ic ools use basic mobile phones and c owd- sou ced da a o deli e low-cos a ming ad ice. These examples highligh ha ugal does no mean in e io a he , i implies echnological adap a ion ha hono s cons ain s while expanding each. Howe e , hese p ojec s o en emain siloed and a e no in eg a ed in o b oade AI- o - De elopmen (AI4D) s a egies o linked o global collabo a ions. 2.5 Iden i ied Gaps and Ra ionale o he P oposed Model Despi e p omising case s udies, se e al c i ical gaps emain in bo h heo y and applica ion: 1.Localiza ion emains na owly de ined. Mos F ugal AI deploymen s ocus on egional language in e aces, ye ail o accoun o cul u al hyb idi y (such as Tamil–F ench–English bilingualism in Puduche y) ha could os e globally scalable models. 2.Diaspo a engagemen is la gely absen . While coun ies like India ha e apped diaspo a o emi ances and philan h opy, hei ole as AIco-inno a o s especially in linguis ically dual o mul icul u al egions— emains unde explo ed. 62 3.F ugal AI is a ely in eg a ed in o o mal go e nance s uc u es. Many solu ions a e NGO-led o s a up-ini ia ed and lack alignmen wi h s a e digi al missions o local adminis a i e sys ems limi ing scale and sus ainabili y. 4.No eplicable ins i u ional amewo k cu en ly exis s o b idge: Global e hical AI manda es (e.g., SDGs, UNESCO) Design-le el inno a ion (e.g., open-sou ce, low-compu e models) Local go e nance (e.g., Digi al India, Sma Ci ies) C oss-cul u al collabo a ion (e.g., Indo–F ench coope a ion) To be e syn hesize he discussion abo e, he ollowing amewo k in eg a es h ee key dimensions—global digi al manda es, ugal AI design p inciples, and Puduche y’s unique socio echnical con ex . This isual model helps concep ualize how scalable, e hical, and a o dable AI can be g ounded in bo h in e na ional de elopmen p io i ies and localized go e nance ecosys ems. I also p o ides he ounda ional logic o he implemen a ion oadmap p oposed la e in his chap e . Sou ce: De eloped by he au ho based on UN SDGs, F ugal Inno a ion Li e a u e, Digi al India guidelines, and Indo–F ench cul u al coope a ion s a egies. Figu e 2.1 Concep ual F amewo k o F ugal AI Deploymen in he Global Sou h This igu e isualizes he con e gence o global digi al public goods, ugal design p inciples, and localized socio-poli ical asse s in Puduche y. I si ua es F ugal AI as he ocal s a egy uni ing in e na ional manda es (like he SDGs), cos -e ec i e AI ools, and Indo–F ench collabo a ion oppo uni ies. 3: Resea ch Me hodology This sec ion ou lines he me hodological amewo k employed o concep ualize and alida e he applica ion o F ugal AI in he Union Te i o y o Puduche y. Gi en he in e disciplina y na u e o he s udy encompassing de elopmen economics, digi al go e nance, AI design, and egional planning a quali a i e, explo a o y esea ch app oach was adop ed, suppo ed by case syn hesis and policy mapping. 3.1 Resea ch Objec i es • To examine how F ugal AI p inciples can be con ex ually applied in public se ices such as educa ion, heal h, and go e nance. 63 • To analyze he socio echnical eadiness o Puduche y o ac as a es bed o F ugal AI inno a ion. • To design a eplicable model ha aligns wi h global policy manda es (e.g., SDGs, UNESCO AI e hics) and local go e nance sys ems. 3.2 Resea ch Design The esea ch d aws upon a mul i-me hod quali a i e amewo k, including: • Documen a y Analysis: Cen al and S a e policy documen s (e.g., Digi al India, NDHM, U.T. Budge ), Indo F ench MoUs, and F ench digi al go e nance epo s. • Case S udy Syn hesis: Compa a i e insigh s om AI- o -de elopmen ini ia i es ac oss India, Kenya, Rwanda, Bangladesh, and F ance. • Theo e ical Ancho ing: G ounded in Roge s’ Di usion o Inno a ions Theo y and Sen’s Capabili y App oach o examine how low-cos AI in e en ions expand human choices in cons ained en i onmen s. 3.3 Da a Sou ces Table 3.1: Da a Sou ces Used in he S udy Da a Type Sou ces Policy F amewo ks Digi al India, NDHM, PM Ga i Shak i, Sma Ci ies Mission, Puduche y S a e Budge Ins i u ional Repo s Alliance F ançaise, F ench Consula e, UNDP, UNESCO AI Guidelines Case S udies Published esea ch on eSanjee ani, DIKSHA, Aap i Ins i u e AI pilo s Local Inpu s (Indi ec ) Adminis a i e bluep in s, budge speeches, Indo F ench uni e si y collabo a ions Sou ce: Compiled by he au ho based on na ional policy po als, ins i u ional documen s, and Indo–F ench academic collabo a ions. 3.4 Analy ical F amewo k The ollowing iadic analysis s uc u e was used: 1. Inpu –P ocess–Ou come Mapping To assess how exis ing o p oposed AI ools ope a e wi hin esou ce-limi ed en i onmen s. 2. Ba ie –Enable Ma ix To iden i y socio-cul u al, echnological, linguis ic, and ins i u ional a iables a ec ing adop ion. 3. Capabili y Enhancemen Lens To examine how F ugal AI con ibu es o expanding agency, access, and empowe men —pa icula ly o ma ginalized popula ions in bilingual egions like Puduche y. 64 3.5. Scope and Limi a ions This is a design-le el s udy, ocused on amewo k building and easibili y modeling, no an empi ical e alua ion. While case syn hesis and seconda y da a p o ide eliabili y, p ima y s akeholde in e iews and on-g ound deploymen a e p oposed as u u e phases (discussed in Sec ion 8). The Indo–F ench lens is uniquely sui ed o Puduche y and may equi e con ex ual adap a ion elsewhe e. Figu e 3.1: Cons uc ion Logic o he F ugal AI Implemen a ion Model o Puduche y The ollowing in og aphic p esen s he laye ed cons uc ion logic o he p oposed F ugal AI model, showcasing how i in eg a es na ional policy amewo ks, global manda es, and Indo F ench collabo a ions h ough g ounded design app oaches and localized il e s. This o ms he basis o he Indo–F ench F ugal AI Inno a ion Hub (IFFAIH), which ancho s he p ac ical applica ions discussed in subsequen sec ions. Sou ce: De eloped by he au ho based on SDGs, Ama ya Sen’s Capabili y App oach, Roge s’ Di usion o Inno a ion Theo y, Digi al India, and Indo– F ench academic and policy collabo a ions. Figu e 3.1: Cons uc ion Logic o he F ugal AI Implemen a ion Model o Puduche y. This isual amewo k in eg a es global e hical manda es, ugal design p inciples, and localized socio-poli ical il e s o es ablish a eplicable and 65 scalable inno a ion hub sui able o esou ce-cons ained con ex s like Puduche y. 4: Case S udies o F ugal AI in Ac ion To alida e he concep ual g ounding o he F ugal AI model p oposed in his chap e , his sec ion analyzes eal-wo ld examples om India and o he coun ies in he Global Sou h. Each case illus a es how low-cos , scalable, and con ex - sensi i e digi al ools ha e been used o o e come in as uc u al, linguis ic, o economic limi a ions in deli e ing public se ices. 4.1 India: eSanjee ani Na ional Telemedicine Pla o m Launched by he Minis y o Heal h and Family Wel a e, eSanjee ani is a low- bandwid h, b owse -based eleconsul a ion pla o m ha suppo s o e 80 million consul a ions. Buil on open-sou ce code, i ope a es e icien ly in semi-u ban and u al se ings wi hou needing high-end in as uc u e. This exempli ies F ugal AI in heal hca e, whe e diagnos ic p o ocols a e semi-au oma ed and mul ilingual sc ip s enable wide each. 4.2 Bangladesh: AI o Flood Fo ecas ing In collabo a ion wi h Google and local agencies, Bangladesh de eloped a ugal AI model o eal- ime lood ale s using sa elli e da a and machine lea ning. The sys em, accessible ia basic SMS, helped educe e acua ion esponse ime by o e 60%. This case illus a es AI o clima e esilience using minimal ech laye s and con ex ual design. 4.3 Rwanda: Ma e nal Heal h AI Cha bo s Th ough he Sma A ica ini ia i e, Rwanda deployed mul ilingual AI cha bo s o assis expec an mo he s wi h p ena al in o ma ion. The bo s ope a e on ea u e phones using oice/ ex hyb ids in local languages. This aligns wi h he Capabili y App oach by expanding agency and knowledge, especially in digi ally unde se ed communi ies. 4.4 Tamil Nadu: DIKSHA Pla o m in Ve nacula EdTech Tamil Nadu’s adap a ion o he na ional DIKSHA pla o m localized open educa ional esou ces in Tamil o go e nmen school eache s. The pla o m uses AI-based ex - o-speech ools and pe o mance dashboa ds, enabling ugal pe sonaliza ion o con en and acking wi hou majo se e in es men . 4.5 Puduche y: Alliance F ançaise Digi al Co-lea ning Hubs Though no ye ully in eg a ed wi h AI, he Alliance F ançaise’s language co- lea ning model in Puduche y o e s a ounda ion o de eloping bilingual, c oss- cul u al AI aining en i onmen s. This highligh s he po en ial o F anco Indian digi al co-inno a ion, whe e ugal pedagogical ools could be enhanced by AI- d i en pe sonaliza ion. Syn hesis These case s udies unde sco e how F ugal AI p inciples ha e been success ully applied ac oss sec o s o en in geog aphies wi h sociolinguis ic complexi y and in as uc u e cons ain s simila o Puduche y. They in o m he implemen a ion 66 logic o he Indo–F ench F ugal AI Inno a ion Hub (IFFAIH) discussed in he ollowing sec ions. To u he consolida e he ele ance and eplicabili y o ugal AI s a egies ac oss di e en socio-economic and cul u al se ings, he ollowing able o e s a compa a i e summa y o no able case s udies om India and o he coun ies in he Global Sou h. Each p ojec demons a es dis inc aspec s o ugal AI design and implemen a ion—such as cos -e iciency, minimal compu e dependency, local language in eg a ion, o pa ne ship-d i en scaling. Thei compa a i e analysis no only alida es he ounda ional p inciples p oposed in his s udy bu also highligh s speci ic insigh s ha can be adap ed o Puduche y’s adminis a i e, cul u al, and geog aphic con ex . Table 4.1: Compa a i e Summa y o F ugal AI Case S udies Coun y/Region P ojec Name Sec o F ugal AI Fea u e Key Rele ance o Puduche y India eSanjee ani Heal hca e Low-bandwid h eleconsul a ion; open-sou ce p o ocols Applicable o PHCs in u al Ka aikal and Yanam Bangladesh AI Flood Ale s Clima e Resilience SMS-based AI o ecas ing; minimal in as uc u e Disas e p epa edness o coas al looding Rwanda Ma e nal Heal h Cha bo s Ma e nal Heal h Voice/ ex - based cha bo s on ea u e phones Ma e nal ca e ou each in mul ilingual illages Tamil Nadu DIKSHA Ve nacula EdTech Educa ion AI TTS ools; low-cos pe o mance dashboa ds Enhances go e nmen school digi al deli e y Puduche y Alliance F ançaise Digi al Hubs Bilingual Lea ning Po en ial o AI co-c ea ion in bilingual se ings Founda ion o Indo– F ench co- inno a ion model Sou ce: Compiled by he au ho om public sec o case epo s, academic publica ions, and Sma A ica, DIKSHA, and Alliance F ançaise esou ces. 67 5: Implemen a ion Model and Design Fea u es D awing om he p eceding li e a u e, case analyses, and con ex ual mapping, his sec ion p esen s he Indo–F ench F ugal AI Inno a ion Hub (IFFAIH) as a scalable and policy-aligned implemen a ion model ailo ed o he Union Te i o y o Puduche y. The model is oo ed in ugaldesign philosophy, guided by e hical AI p inciples, and suppo ed h ough bila e al coope a ion be ween India and F ance. 5.1 Pilla s o he IFFAIH Implemen a ion Model The IFFAIH model is s uc u ed a ound i e in e dependen pilla s: Table 5.2: Co e Pilla s o he IFFAIH Implemen a ion Model Pilla Desc ip ion 1. Policy Con e gence In eg a es SDG a ge s, Digi al India, NDHM, and Indo–F ench MoUs 2. F ugal Design Logic Low-cos , low-compu e, high-access ools; p omo es open-sou ce and modula AI 3. Cul u al Localisa ion Mul ilingual (Tamil–F ench–English) in e aces and communi y-cen ic aining 4. Ins i u ional In eg a ion Ancho ed wi hin Puduche y colleges, PHCs, and Alliance F ançaise ne wo ks 5. Sus ainabili y Engine Embedded in S a e budge s and CSR channels; moni o ed ia open me ics Sou ce: Cons uc ed by he au ho based on analysis o policy con e gence, ugal design li e a u e, and Indo–F ench coope a ion amewo ks. 5.2 Key Design Fea u es 1. Bilingual Accessibili y In e aces a e designed o wo k in Tamil and F ench, enhancing digi al pa icipa ion and p omo ing equi y in se ice access. 2. De ice-Agnos ic Deploymen Solu ions a e designed o unc ion ac oss basic sma phones, able s, and communi y desk ops ideal o u al o unde se ed pocke s. 3. Modula Mic o se ices A chi ec u e Encou ages in e ope abili y wi h heal h, educa ion, and go e nance pla o ms al eady ope a ional unde Digi al India. 4. Fede a ed Lea ning P inciples Ensu es p i acy-p ese ing AI aining on decen alized da a—aligned wi h India’s da a so e eign y goals and GDPR-like amewo ks in F ance. 5. Public P i a e Academic Pa ne ships (PPAP) Le e ages F ance’s AI esea ch ecosys em (e.g., INRIA, CNRS) and India’s g ass oo s inno a o s o co-design and co-implemen a ion. 74 ad ances a eplicable ision o digi al equi y. Th ough a ge ed pilo ac i i ies, igo ous moni o ing, and long- e m sus ainabili y planning, he model ensu es ha a i icial in elligence becomes an inclusi e enable no an eli e dis up o . Ul ima ely, he chap e a gues ha e hical, low-cos , and communi y-aligned AI is no a dis an ideal. Wi h p ope ins i u ional commi men and global-local pa ne ships, i is a s a egy eady o immedia e, scalable, and measu able deploymen beginning wi h Puduche y, and expanding whe e e inclusion is a p io i y. Re e ences 1. Aap i Ins i u e. (2022).AI and da a s ewa dship pilo s. h ps://aap i.in 2. Alliance F ançaise. (2024).F ench language and cul u al ne wo k in India. h ps://a india.o g 3. Campus F ance. (2024).S udy and esea ch in F ance. h ps://www.campus ance.o g 4. DIKSHA. (2023).Digi al in as uc u e o knowledge sha ing. h ps://diksha.go .in 5. Digi al India. (2023).Digi al India P og amme. h ps://digi alindia.go .in 6. F ench Embassy in India. (2023).Indo-F ench coope a ion in educa ion and esea ch. h ps://in.amba ance.o g 7. Ghosh, S. (2021).F ugal inno a ion and de elopmen . Jou nal o Inno a ion and En ep eneu ship, 10(1), 1–14. 8. Kuma , R., & Das, P. (2022).Low-cos AI applica ions in heal h and educa ion. AI & Socie y,37(2), 187–204. 9. La F ench Tech. (2023).F ench s a ups and inno a ion ecosys em. h ps://la ench ech.com 10. NITI Aayog. (2021).Responsible AI o All: Na ional s a egy o a i icial in elligence. h ps://ni i.go .in 11. Na ional Heal h Au ho i y. (2023).Na ional Digi al Heal h Mission (NDHM). h ps://ndhm.go .in 12. PM Ga i Shak i. (2023).Na ional Mas e Plan o Mul i-modal Connec i i y. h ps://pmga i.go .in 13. Puduche y Planning Depa men . (2024).Budge speeches and digi al policy no es. h ps://py.go .in 14. Roge s, E. M. (2003).Di usion o inno a ions (5 h ed.). F ee P ess. 15. Sen, A. (1999).De elopmen as eedom. Ox o d Uni e si y P ess. 16. Sma Ci ies Mission. (2024).Minis y o Housing and U ban A ai s. h ps://sma ci ies.go .in 17. UNESCO. (2021).Recommenda ion on he e hics o a i icial in elligence. h ps://unesdoc.unesco.o g 18. eSanjee ani. (2023).Telemedicine se ice by Minis y o Heal h. h ps://esanjee ani.in