Transcript The Future of AI is Here Teck Chia Partner, Exponent.vc Why talk

The Future of AI is
Here
Teck Chia
Partner, Exponent.vc
Why talk about AI
Foundational shift in computing
Transcends markets, culture, geography
Early but impactful
Structure
Quick overview of AI technology
Survey of AI activities
Opportunities in AI
Q&A
What is AI?
Humans
Natural Intelligence is using the brain to solve
problems by learning and reasoning from
information.
What is AI?
Machines
Artificial Intelligence is using algorithms to solve
problems by learning and reasoning from
information.
Types of AI
Narrow Artificial Intelligence
(Weak AI)
Artificial General Intelligence
(Strong AI)
Narrow Artificial Intelligence
(Weak AI)
Single Domain
Not easily transferrable
Examples: Self-driving cars, Apple Siri, Spam filters
Artificial General Intelligence
(Strong AI)
Can learn anything
Learn how to learn
Examples: Only in science fiction & movies
(for now)
Machine Learning
Class of algorithms that can learn from data or
information to solve problems
Most of today’s AI is using ML techniques
Machine Learning
Paradigms
Supervised Learning
Unsupervised Learning
Reinforcement Learning
Supervised Learning
Apple
Apple
Apple
Apple
Apple
Supervised
Learning
Algorithm
Supervised Learning
???
???
Trained
Supervised
Learning
Model
Apple
Not Apple
Supervised Learning
Image recognition
Speech recognition
Predictive Analytics for businesses
Unsupervised Learning
Unsupervised
Learning
Algorithm
Unsupervised Learning
News or content clustering
Recommendation systems
Anomaly detection
Reinforcement Learning
Reinforcement Learning
Robotics/Self-driving
Game playing (AlphaGo)
Quant trading
What about Deep Learning?
AKA Neural Networks
Used in all ML paradigms
≈
Why AI is exciting
today
Compute Power
Increase in power, lower in cost: GPUs, ASICs
Abundance of Data
Sensors are everywhere
Consumer behaviors shifting online
Big data infra available and cheap
Access to ML Tools
What’s Happening in
AI Today
Big Company AI
Vision, Speech, Natural Language, Video, Social
Robotics, Infrastructure, Tools, Hardware, Data
Startups
Mostly focused on
narrow AI in vertical markets
(Small number of them attacking big AI problems)
Self-driving
Healthcare
Drug discovery
Medical imaging and diagnostics
Analytics & risk management
Medical research
Health monitoring/Wearables
Software & DevOps
Site reliability and monitoring
Software engineering
Testing Automation
Fintech
Quant trading bots
Personal finance management
Credit & risk assessment
Enterprise Ops
Sales Optimization
Marketing automation & ads targeting
Supply chain and logistics
Security
Fraud prevention
Malware & spam detection
Malicious attack detection
Robotics
Food-making robots
Security patrol robots
Autonomous drones
Industrial robots
(Self-driving cars)
Creativity & Fun
Face recognition & filters
AI-generated music & art
AI computer player
Artificial General Intelligence
Turing test
Learn anything (including how to learn)
Moravec’s Paradox
Opportunities
Verticals where you can acquire lots of training
data
Tools & utilities to make ML easier to use
Hardware & infrastructure for ML
Q&A