Alla organisationer kommer vara AI organisationer
0 likes137 views
Keynote fr奪n Offentliga Rummet (SKR) 25 feb 2021
Alla organisationer kommer vara AI organisationer
- 1. Alla organisationer kommer att vara AI-organisationer Kye Andersson / Major Impact Initiatives, AI Sweden Feb / 2021
- 2. Accelerate the use of AI for the benefit of our society, our competitiveness, and for everyone living in Sweden Mission statement for AI Sweden
- 3. Avskriva utmaningen som t.ex teknisk, och l奪ta n奪gon annan utforska den. D奪 slipper vi f旦r辰ndring. F旦r n辰stan alla m辰nniskor, d奪 det utmanar v奪r identitet. F旦r辰ndring 辰r obekv辰mt 1 Ledande roller i organisationer som vilar p奪 erfarenhet har sv奪rare att blotta okunskap. S辰rskilt obekv辰mt f旦r ledare 2 Vanlig skyddsmekanism det d辰r g辰ller inte mig 4 Medvetenhet kring v奪ra skyddsmekanismer g旦r oss mer mottagliga f旦r f旦r辰ndring Medvetenhet f旦der 旦ppenhet 5 Vi reagerar instinktivt med skepsis. Det sker omedvetet, f旦r att skydda oss sj辰lva. Skyddsmekanismer 3 Nu pratar vi om AI och framtiden Nu 辰r vi lite medvetna om v奪ra skyddsmekanismer 6 En kort check-in, innan vi pratar om AI
- 4. A lot of talk about it. Very little understanding. AI
- 6. Verkligheten
- 7. The timeline for when our biosphere no longer sustains human life, with our current trajectory, is shorter than all predictions around when General Artificial Intelligence occurs. Vi har mer akuta problem 辰n m旦rdarrobotar. AI kan hj辰lpa oss l旦sa dem.
- 8. Vad inneb辰r d奪 AI?
- 9. Sundar Pichai / CEO Alphabet Google AI 辰r en av de viktigaste sakerna m辰nskligheten arbetar med. Det 辰r st旦rre 辰n, jag vet inte, elektriciteten eller elden.
- 10. Aff辰rsid辿erna f旦r n辰sta 10,000 startups 辰r l辰tta att f旦rutsp奪; Ta X och l辰gg till AI Kevin Kelly / Founder Wired
- 11. Bernard J Tyson / Kaiser Permanente Ingen l辰kare borde idag praktisera utan st旦d av AI i sin verksamhet. Utan det 辰r det helt enkelt om旦jligt att 旦vervaka h辰lsa utifr奪n ett helhetsperspektiv
- 12. Peltarion - the operational AI platform Teknisk utveckling utifr奪n ett ekonomiskt perspektiv Internet Kostnadsminskning f旦r Distribution av information AI 1950 1960 1970 1980 1990 2000 2010 Transistorer Kostnadsminskning f旦r Matematiska ber辰kningar Kostnadsminskning f旦r Intelligens & Prediktioner 2020
- 13. Peltarion - the operational AI platform Artificiell intelligens har varit med l辰nge Use data to automatically learn to make predictions Learn to both represent data and make predictions 1950 1960 1970 1980 1990 2000 2010 Artificial Intelligence Science and engineering of building intelligent machines Machine Learning Deep Learning
- 14. Exponentiell utveckling 2007 2017 Storlek neurala n辰t (noder) 10油000 1油000油000油000 Storlek dataset (t.ex. bilder) 1000 100油000油000
- 15. Tack vare gamers
- 16. 2017 vinner OpenAI 旦ver v辰rldsm辰staren i Dota
- 17. 2017 vinner AlphaGo 旦ver v辰rldsm辰staren i Go
- 18. Peltarion - the operational AI platform Artificiell intelligens har varit med l辰nge Use data to automatically learn to make predictions Learn to both represent data and make predictions 1950 1960 1970 1980 1990 2000 2010 Artificial Intelligence Science and engineering of building intelligent machines Machine Learning Deep Learning
- 19. Artificiella Neurala N辰tverk Kan utifr奪n data, utan regler, hitta samband och dra precisa slutsatser
- 20. De senaste AI-teknikerna g旦r att vi kan l旦sa ol旦sliga problem Problem eller uppgifter som tidigare kr辰vt f旦r mycket resurser eller tid.
- 21. AI anv辰nds redan i tj辰nster i v奪r vardag
- 26. AI kan hj辰lpa oss i alla omr奪den i samh辰llet
- 27. Amplitude Benign Malignant Uppt辰cka hudcancer, astma, allergier & eksem. AI & Electronic Spectroscopy
- 28. Screening f旦r br旦stcancer via smartphone med infrar旦d kamera. Alla 奪ldrar. Sjuksk旦terskor. Ingen ber旦ring.
- 29. Livsl辰ngd, laddning, urladdning, balansering. Optimering av batterier. B辰ttre r辰ckvidd.
- 30. IMAGE THAT SUPPORTS THE HEADLINE GOES HERE V辰rdering av fastigheter, hus och l辰genheter
- 31. Kombination av olika sorters data 50% 旦kning i kvalitet av prediktioner SATELLITBILDER GATUBILDER RITNINGAR ANNONSTEXTER MKLARBILDER TERRNGKARTOR TOPOGRAFI
- 32. Fr奪n 3 timmar H奪rdvarukostnad: 300MSEK Felmarginal: 14% To 100 ms H奪rdvarukostnad: 100TSEK Felmarginal 5% Balansera enegergin辰t genom b辰ttre v辰der- prediktioner IMAGE THAT SUPPORTS THE HEADLINE GOES HERE
- 33. V辰derstationer Vindturbiner Topogra鍖 V辰derprognoser Satelliter VISIBLE IR VAPOUR Kraftn辰tet Fr奪n 14% till 5%* *Felmarginal
- 34. Fr奪n Kr辰ver radiologer och fokuserat, tidskr辰vande manuell analys av bilder. Till Segmentering p奪 millisekunder Resultat i niv奪 med de b辰sta radiologerna i v辰rlden Hitta och segmentera hj辰rntum旦rer Flair T1C True tumor T1 T2 Predicted tumor
- 35. Insulin "prediktioner" upp till 12 timmar in I framtiden; samt relevanta beteenderekommendationer 91% av prediktioner inom +/- 50 mg/dL av glukosmeter-m辰tv辰rden och 75% inom +/- 27 mg/dL Prediktion av insulinniv奪er + r奪d och st旦d
- 36. From Rekommendationer 辰r aff辰rskritiska f旦r musiktj辰nster. Traditionellt manuell kurering med begr辰nsningar och subjektiv smak. To Tr辰ffs辰ker klustring av musik m旦jligg旦r tips av musik som du gillar i h旦gre utstr辰ckning. Musik som du faktiskt gillar.
- 37. Fr奪n Energif旦rbrukning enligt f旦rinst辰llda regler. Till Realtidsjustering och f旦ruts辰gelser baserat p奪 anv辰ndning av lokaler och m辰nskliga behov, samt yttre faktorer. Optimera anv辰ndning av energi IMAGE THAT SUPPORTS THE HEADLINE GOES HERE
- 38. Fr奪n Individuella analyser som tar l奪ng tid. Till Aggregerad och tillg辰ngliga insikter f旦r att skapa styrmedel f旦r energieffektivisering. Optimera anv辰ndning av energi, makroniv奪 IMAGE THAT SUPPORTS THE HEADLINE GOES HERE
- 39. kande f旦rm奪ga hos AI att detektera h辰lsoavvikelser med hj辰lp av HRV. 95.1% tillf旦rlitlighet f旦r uppt辰ckt av diabetes med hj辰lp av enbart HRV. Deep Learning & Heart Rate Variability (HRV)
- 40. Hj辰lpa ALLA barn n奪 sin fulla potential i skolan Assistera l辰rare med skr辰ddarsydda inl辰rningsplaner baserade p奪 varje individs inl辰rningsbehov, vid varje given tidpunkt.
- 41. Identifying content relationships through NLP and interaction created knowledge graphs Knowledge Graph The worlds leading AI technology for learning 5 Pinpointing knowledge levels & gaps in 75% less time through AI-powered adaptive assessment Adaptive Assessment Improving learning outcomes up to 50% by using AI to close every learners knowledge gaps Adaptive Learning Paths Improving knowledge retention by up to 3x through AI-powered models of memory and repetition Knowledge Retention
- 43. Peltarion - The Operational AI Platform BERT
- 44. Ett enormt genombrott 2019. AI som kan f旦rst奪r stora textm辰ngder. Hitta liknande texter, som betyder samma sak, har samma budskap, tonalitet/tonl辰ge, avs辰ndare, m.m.
- 45. Paris Hilton stays with Orlando Bloom Orlando stays at the Hilton in Paris
- 46. Titta p奪 offertf旦rfr奪gningar eller offerter, automatisk analys och j辰mf旦relser med historisk data. S旦ka igenom 100000-tals dokument efter mest relevant information. IMAGE THAT SUPPORTS THE HEADLINE GOES HERE Automatiserad upphandling och/eller offerering.
- 47. IMAGE THAT SUPPORTS THE HEADLINE GOES HERE F旦rslag p奪 diagnoser baserat p奪 patientsamtal + andra datapunkter. Litteratur- studier i realtid.
- 48. IMAGE THAT SUPPORTS THE HEADLINE GOES HERE Identifiera behov av st旦d i stora komplexa organisationer St旦dja f旦r辰ndring Hitta insikter bortom vedertagna sanningar, etablerad kultur och otrygghet, i stora organisationer med 1000-tals anst辰llda.
- 49. I alla organisationer spenderar de flesta m辰nniskor tid och energi p奪 hur deras kollegor uppfattar dem. Det kallas internpolitik. Det hindrar organisationer, och alla som arbetar i dem, fr奪n att n奪 sin fulla potential och forts辰tta utvecklas. Robert Kegan (American developmental psychologist)
- 50. IMAGE THAT SUPPORTS THE HEADLINE GOES HERE Var beh旦ver vi st旦d och hj辰lp f旦r att fungera och m奪 b辰ttre? Objektiv data & transparenta insikter som st旦d till ledare och hela verksamheten.
- 51. De som operationaliserar AI kommer vara de organisationer d辰r m辰nniskor kan n奪 sin fulla potential Fokus p奪 v辰rde Applicera AI f旦r att skapa v辰rde verallt I alla delar av verksamheten
- 52. AI 辰r INTE fr辰mst en teknisk utmaning
- 53. Oscar H旦glund / CEO Epidemic Sound Som ledare m奪ste du ha mod och v奪ga anv辰nda AI som st旦d f旦r din k辰rnaff辰r. Det kommer g旦ra er minst 10x b辰ttre. Avf辰rda inte AI som en teknisk fr奪ga eller som forskning.
- 54. Den m辰nskliga faktorn 辰r det st旦rsta hindret
- 55. F旦r辰ndring kommer aldrig mer ske s奪 h辰r l奪ngsamt
- 56. ngkraften 150 奪r
- 60. 2020-2030 vinnare kommer vara de som nyttjar AI f旦r att skapa Bortom m辰nsklig f旦rm奪ga Snabbhet Fler kanaler kad tillg辰nglighet Individanpassning Kundfokus St辰ndig f旦r辰ndring nnu fler medium Proaktivitet Nya insikter 1990-2020 vinnare var de som omfamnade internet & mobilt, vilket gav
- 61. A lot of talk about it. Very little understanding. AI
- 62. Artificiella Neurala N辰tverk Kan utifr奪n data, utan regler, hitta samband och dra precisa slutsatser
- 63. N奪gra f奪 satsar enormt mycket! Utmaningar f旦r demokrati och fri marknadsekonomi.
- 64. Peltarion - the operational AI platform Djupa neurala n辰t Traditionell maskininl辰rning Resultat M辰ngd data Storlek och tillg奪ng till data spelar roll
- 66. The potential to double economic growth rates 1 2 3 3 4 US May June July Baseline AI steady state GVA, % Growth (Gross Value Added) Source: Accenture and Frontier Economics UK SWE GER USA
- 67. K辰nsliga tittare varnas! F旦ljande slide inneh奪ller obekv辰m sanning.
- 68. USA och Kina investerar mer i AI 辰n resten av v辰rlden - industri, akademin och offentlig sektor Vi 辰r inte b辰st i v辰rlden, men vi 辰r topp 20.
- 69. Europa t辰vlar inte ens om bronsmedalj i AI Kai-Fu Lee / Ex president Google China
- 70. Utmanar v奪r identitet. Uppfattas som potentiell maktf旦rlust och hot mot v奪r trygghet. F旦r辰ndring 辰r obekv辰mt 1 M辰nniskor i ledande st辰llning med stort ansvar har oftast sv奪rast att blotta okunskap och svagheter. S辰rskilt obekv辰mt f旦r ledare 2 Avf辰rda som teknisk utmaning. Prata etik. Odemokratiskt. V奪r egen v辰g. Skyddsmekanismer 3
- 73. Vid middagsbordet brukade pappa uppmana mig och min bror att dela dagens misslyckanden. Ist辰llet f旦r att bli besviken eller uppr旦rd s奪 firade han att vi hade f旦rs旦kt. Sara Blakey / Founder of Spanx
- 74. Michael Jordan / Legend Jag har missat mer 辰n 9000 skott i min karri辰r. Jag har f旦rlorat n辰stan 300 matcher. 26 g奪nger har jag f奪tt f旦rtroendet att ta ett matchavg旦rande skott och missat. Jag har misslyckats om och om igen i mitt liv. Det 辰r d辰rf旦r jag lyckas!
- 75. Elon Musk / Space X, Tesla, etc T辰nk alltid p奪 hur du kan g旦ra saker b辰ttre och ifr奪gas辰tt dig sj辰lv
- 76. 01/ Basic understanding of AI 03/ Becoming a data driven organization 04/ Mastery of data 05/ AI Expertise 06/ Tooling suited for business 02/ Leadership & organization Source: Tv奪 omr奪den allra viktigast
- 77. The 45 minute AI introduction for leaders Elements of AI bit.ly/45minAI elementsofai.com
- 78. Coming soon
- 79. The rest of this document outlines critical areas that will affect your progress towards getting full value out of AI. Your organisation can access the full value potential of AI when: The CEO is responsible for, and has an inventory of, the business opportunities that AI offers for the organisation and the industry. Clearly connected to the organisations vision and mission statement. Valuation of each opportunity Effort assessment List of needed requirements to realize each opportunity The CXO team regularly takes business critical decisions based on patterns and insights generated from data with the help of AI. Patterns and predictions beyond what human analysis, or experience, could have found. Patterns across multiple areas of operation, using data from across the organisation. Work in progress, not final copy or layout. Din organisation kan nyttja AI fullt ut n辰r
- 80. Clearly connected to the organisations vision and mission statement. Valuation of each opportunity Effort assessment List of needed requirements to realize each opportunity The CXO team regularly takes business critical decisions based on patterns and insights generated from data with the help of AI. Patterns and predictions beyond what human analysis, or experience, could have found. Patterns across multiple areas of operation, using data from across the organisation. Without necessarily understanding how (the logic process) these predictions were reached. Your organisation is clearly structured to leverage AI Every CXO understands AI and its opportunities for your organisation Every CXO has a clear plan for how AI will increase the output of their area of responsibility. Work in progress, not final copy or layout. Din organisation kan nyttja AI fullt ut n辰r
- 81. generated from data with the help of AI. Patterns and predictions beyond what human analysis, or experience, could have found. Patterns across multiple areas of operation, using data from across the organisation. Without necessarily understanding how (the logic process) these predictions were reached. Your organisation is clearly structured to leverage AI Every CXO understands AI and its opportunities for your organisation Every CXO has a clear plan for how AI will increase the output of their area of responsibility. The organisation has deployed operational AI models in business critical functions, that augment the current value proposition, business results, innovation ability, speed to market and problem solving ability. Work in progress, not final copy or layout. Din organisation kan nyttja AI fullt ut n辰r
- 82. Without necessarily understanding how (the logic process) these predictions were reached. Your organisation is clearly structured to leverage AI Every CXO understands AI and its opportunities for your organisation Every CXO has a clear plan for how AI will increase the output of their area of responsibility. The organisation has deployed operational AI models in business critical functions, that augment the current value proposition, business results, innovation ability, speed to market and problem solving ability. Work in progress, not final copy or layout. Din organisation kan nyttja AI fullt ut n辰r
- 83. Work in progress, not final copy or layout. Betygskort f旦r m辰tning av f旦r辰ndring
- 84. Your organisation can access the full value potential of AI when: The CEO is responsible for, and has an inventory of, the business opportunities that AI offers for the organisation and the industry. Clearly connected to the organisations vision and mission statement. Valuation of each opportunity Effort assessment List of needed requirements to realize each opportunity The CXO team regularly takes business critical decisions based on patterns and insights generated from data with the help of AI. Patterns and predictions beyond what human analysis, or experience, could have found. Patterns across multiple areas of operation, using data from across the organisation. Without necessarily understanding how (the logic process) these predictions were reached. Your organisation is clearly structured to leverage AI Every CXO understands AI and its opportunities for your organisation Every CXO has a clear plan for how AI will increase the output of their area of responsibility. The organisation has deployed operational AI models in business critical functions, that augment the current value proposition, business results, innovation ability, speed to market and problem solving ability. Work in progress, not final copy or layout.
- 85. You should set goals beyond your reach so you always have something to live for. - Ted Turner * att vara b辰st" 辰r inget m奪l
- 86. President Barack Obama Nordic nations punch above their weight
- 87. Peltarion - the operational AI platform Djupa neurala n辰t Traditionell maskininl辰rning Resultat M辰ngd data Storlek och tillg奪ng till data spelar roll
- 88. Nyckeln till framg奪ng Nyfikenhet, f旦r辰ndringsvilja och samarbete
- 89. Accelerating applied AI in Sweden
- 90. Accelerate the use of AI for the benefit of our society, our competitiveness, and for everyone living in Sweden Mission statement for AI Sweden
- 91. National center for applied AI Neutral, non-profit and broadly funded Here to help AI help you
- 93. Networks National and international Collaboration Share knowledge, data, best practice Excellence Nationally & internationally building on regional strengths Leadership Mindset and organization are key factors How we create impact
- 94. Training Program Networks Project Portfolio Strategic Partner Program Data Factory Data labs Major Impact Initiatives Applied AI projects Challenges and short projects National and international network Co-locations Reference and focus groups Events and matchmaking AI Startup Landscape AI Change Agent Program AI for Leaders Elements of AI Deep Dive Sessions Datasets Compute and storage capacity Infrastructure testbed Legal Expert Group Data and knowledge sharing Executive AI Accelerator Regional network Startup program Partner offering
- 95. St辰lla om till informationsdriven, AI-redo, individanpassad v奪rd f旦r alla i Sverige. AI-f旦rberedelse Traditionell sjukv奪rd Skapa digitala lager AI transformation Sammanh奪llen v奪rdkedja som nyttjar all tillg辰nglig data f旦r friskv奪rd. SYMPTOMBEHANDLANDE FULLT HLSOSTDJANDE Tillg奪ng till data AI-redo En avg旦rande satsning f旦r framtiden, med stor samverkan mellan ett st辰ndigt v辰xande antal akt旦rer 旦ver hela Sverige. BTTRE & SNABBARE VRD. PREVENTIV FRISKVRD OCH FRISKARE MEDBORGARE. MINSKAD ADMINISTRATION, FRBTTRAD ARBETSMILJ ENORMA KOSTNADSBESPARINGAR. 1 AV21 REGIONER HAR REALISERAT 1% AV SIN AI-POTENTIAL OCH REDAN GJORT STORA BESPARINGAR
- 96. Nu 辰r det dags f旦r er kye.andersson@ai.se
- 97. AI kan hj辰lpa oss att snabbare hitta b辰ttre svar, f旦ruts辰ga saker mer precist, automatisera och individanpassa l旦sningar i st旦rre utstr辰ckning, och mycket annat. I en dr旦mv辰rld, vad skulle du 旦nska att AI anv辰ndes till i framtiden? (f旦rs旦k att inte ta h辰nsyn till begr辰nsningar) I ditt eget liv och vardag? P奪 din arbetsplats? I samh辰llet? I andra omr奪den? Det h辰r 辰r anledningen till att vi alla finns och varf旦r v奪ra organisationer existerar f旦r att bygga en b辰ttre v辰rld f旦r kommande generationer. D辰rf旦r 辰r denna bild med.