Science Should Very Importantly Have: What & How
Introduction to AI
- Generative AI, Challenges and Way Forward
- Attributes of AI - That They Cannot Do but Qualifies as AI, this needs to be taught to them
- Learning
- Decision / Reason
- Self-Correction & Self-Awareness
Difference between AI and generative AI
- AI has Intelligence but Generative AI is something that can Generate/Produce
- The output of AI will be generally a Number or a yes or no decision but generative AI with answers gives suggestions
Training in AI
- Data can be Structured
which is categorized and structured
- Data can be Unstructured
which is NOT categorized and structured
what is Generative AI
Generative AI does it create something New → it does mix & match but it is very useful
AI →
Machine Learning (Machine Learns without Human Interference)→ Deep Learning (Based on Artificial Neural Networks; Tries to Simulate the Human Brain)→
Generative (Capacity to Create Something) & Discriminative (Yes & No)
LLM → Large Language Models
OCR → Optical Character Reading
NLP → Natural Language Processing
Super AI → Combine Quantum & AI
Strong AI → Can Emulate Whole Human Intelligence. There is No Example as of now.
Weak AI → Can Handle only a part of Human Intelligence, All the AI we have right now, is weak AI
The idea is to make AI that is Responsible
Strength of AI → Pattern Recognition Capabilities
Question
AI has the capacity to transform the digital ecosystem. in this context identify the applications of Artificial Intelligence
Applications of AI
- Health
- Google and Apollo have signed an agreement wherein they are using Deep Learning to interpret X-Ray
- It can predict the spread of disease
- Disease Diagnosis
- Suggesting & Creating Medicines & Vaccines
- Research & Development
- Examples
3 Nethra (Microsoft & Forus) → Opthalmology Practise Analysis by this App
- Agriculture from Handout Also
- Can take input from Soil and plants and can suggest plantation techniques
- Can identify and measure the spread of disease
- Credit Wohtiness of Farmers
- E NAM
- Cropping Patterns
- Disease Surveillance & Monitoring
- GM Crops
- AI-based sensors and drones for surveillance and management
- Sowing via Machines
- Dynamic Prices Suggestions, Predicting Demands
- Agriculture in AI on Marines and Fisheries
- Education
- Translation
- Tele Education & Getting People back on diction Ex: Duolingo
- we can provide customized teaching
- An example is X Prep (Customised Study Schedule)& DIKSHA Portal
- Business
- Supply chain and Inventory Management
- Dynamic Pricing
- Targetted and Personalised Ads
- Managing Crises with real-time data for decision-making and suggestions
- Inventory Management
- Leakage Management
- Security in General
- Image Processing
- Before Conflict Tracking Hate Speeches and Internet Heat
- communication calls
- can identify hate speech
- Example: Snowden Debate
- Market
- Capital Market Prediction
- Analysing Market Trends
- Facial Recognition
- Digi Yatra
- Quantum Computing
Follow the Life Cycle Approach
- Seed, Soil, Growth, Harvesting, Pos Harvesting, Marketing
- Always Follow Life Cycle Approach for Education
- Identify the Segments First
Home Work: Applications of Generative AI
Questions: What are the challenges surrounding AI
Ethical Challenges
- Accountability & Legality
Technology Outpaces Regulation
- Algorithm Bias - Both at the time of creation and learning
- Data Bias
- Explainability - ANN’s are very complex, Sometimes it cannot explain its process and output. Essence is that the Decision Making Process cannot be explained by AI
- Informed Consent - The Dilemma is that AI cannot learn if you dont share data without learning
- Threat for Privacy
- We also dont want privacy & we accept terms and conditions without reading
- Deep Fakes
- It can profile us in both Individual Profiling and Societal Profiling to predict our needs wants and decisions
- Prominent Threat of Job Loss
- Malafide Misuse
General Challenges (From Handout)
- Money →
- Gross Expenditure on Research & Development
- In Case of India Pvt Expenditure of R&D is 36% of Total which is very low
- of the total govt spending only 7% of total goes to the higher institutions
Graph on R&D from Handout
taken from NITI at 75
The total is Low, Pvt is Low but the Govt Contribution is also decreasing
This Trend is not good
- Infrastructure, Institution & HRD→
- As per QS World Rankings out of 200 Top Institutions only 3 are in India
- of the total patents filled, indian patent rank is at 10, israel outranks us
- UNESCO school of statistics Report 2019, Plots no of Researchers per million
- For USA - 4200
- China - 1200
- India - 156
- R&D & Innovation →
- Section 3 D & Section 56 → Deals with Evergreening and Compulsory Licensing
- Commercialisation →
Governance Policies on AI
- Policy
- Open Science Policy - Democratise Science for All
- SIT Observatory
- TKS - Traditional Knowledge System
S&T & Innovation Policy STIP
- Institutions
- Under MoST
- → DoSP → All Definitions Roadmaps, FInance and Managements
- → DSIR - CSIR → Research
- DBT
- MeITY
- CDAC - Hardware Maker - Manufacturing
- NIC - National Information Centre
- National Knowledge Network (NKN)
- You create a network in India and then connect it to a Super Computer
- HRD - Human Resources
- Women - KIRAN (Handling Career Gaps), CURIE (Women Specific Institutes),
- Youth - INSPIRE & MANAK Awards
- Disabled & Elderly - TIDE - Technology Interventions for Disabled & Elderly
- R&D & Innovations
- CoE - Centre of Excellence
- TIH - Technology Innovation Hubs
- NIDHI - National Initiative for Developing Harnessing Innovation
- AIM - Atal Innovation Mission
- Other Works
- PRISM by Samsung - Connects Industry and Academy
- Science & Engineering Research Board (SERB) + Intel Deep Learning
- SIH by Govt of India
AIRAWAT - AI Super Computer Cloud has been built by CDAC
Other Concepts
GAN - Generative Adversary Network → DALL E
AI in Infrastructure
- Planning & Design Suggestion
- Material Choice
- Structural Analysis
- Disaster Proof
GPT 4.0
LLM - Large Language Model