Ai current use-and-future-opportunities
1 like198 views
The document discusses artificial intelligence and its applications. It outlines different paradigms for AI, from early expert systems using predetermined rules, to modern machine learning models that learn from data, to future human-machine intelligence where humans help build AI. It also discusses using AI for automation, personalization, smart devices, and education. The role of humans in training models and providing feedback is emphasized.
Ai current use-and-future-opportunities
- 1. CW IN CAPGEMINI WEEK OF INNOVATION NETWORKS 1. Artificial Intelligence
- 2. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 2Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. Before talking about Artificial Intelligence… Do You believe we can automate this process? How about when we feel smart? How practically we can “connect the dots”?
- 3. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 3 Customer Experience Automation Personalization @ Scale UX- UI Smart Device Robotics RPA Chat bots or Voice bots Natural Language Generation Semantic Technology Biometrics Deep Learning Speech Recognition Image and Video analysis Machine Learning …and what then about AI
- 4. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 4 …education to leverage Artificial Intelligence? « …Teachers, leave them kids alone…» …Rules… Experiences…
- 5. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 5 …and the role of Humans? …maybe You remember the Mechanical Turk… …and maybe You get to know the Digital Version of that… …The Human API… …and the Train @ scale…
- 6. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 6 “Old” Paradigm Programming Expert system that simulates the human decision process basing the decision on predetermined rules by humans. Recent Past Paradigm Machine Learning Machine learning can keep HIT in tree ways: 1. Explaining what data sources are responsible for the final decision; 2. Training and Tuning the model; 3. Giving the last word to humans. Machine learns rules from data …Human-in-the-loop (HIT)… On the way Paradigm Human-Machine Intelligence 1. Collect unavailable data by crowdworking 2. Choose best neural network architecture; 3. Reinforcement learning is performed by feedback given from data or human. Human build human intelligence into machine learning process. We are here
- 7. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 7 Programming Machine Learning …to Train the artificial intelligence solution… Human-Machine Intelligence Collect data Validate data Manual data generation Intelligent data generation Generative data by Neural Network Insert rules Into expert system
- 8. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. …keeping in mind Utility Functions…the Ys and its ethics
- 9. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 9 …riding on AI on the Edge …Machine talking to machines… …to substitute traditional Software coding………. ..Your own Personal Assistant… …Explicability… …DoE…the DNA of everything
- 10. Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. 10Presentation Title | Author | Date © 2018 Capgemini. All rights reserved. Contacts Massimo Ippoliti Insights & Data Practice Director massimo.ippoliti@capgemini.com