3. 3
What is artificial intelligence?
Artificial intelligence
is the simulation of
human intelligence by
machines
Machine learning lets
us give computers the
ability to learn without
being explicitly
programmed
Deep learning is a
problem solving
approach for
implementing machine
learning (i.e. neural
networks)
WHAT IS AI?
6. 6
In 1956, researchers gathered at Dartmouth with goal of
giving computers ability to:
1. Reason
2. Understand the world and objects within it
3. Navigate through the world
4. Process, understand, and communicate in natural language
5. Perceive the world around them
6. Develop “generalized intelligence”
The Beginning
THE ORIGINS
https://a16z.com/2016/06/10/ai-deep-learning-machines/
10. 10
Like relational databases… machine learning is a
building block that will be part of everything,
making many things better and enabling some new
and surprising companies and products.
— Benedict Evans, Partner at Andreesen Horowitz
“
WHY AI MATTERS
11. 11
I think it’s gigantic… it’s probably hard to
overstate how big of an impact it’s going to have
on society over the next 20 years.
— Jeff Bezos, CEO of Amazon
“
WHY AI MATTERS
12. 12
The Potential Economic Impact of AI
WHY AI MATTERS
https://www.accenture.com/ca-en/insight-artificial-intelligence-future-growth
13. 13
The Potential Productivity Impact of AI
WHY AI MATTERS
http://www.mckinsey.com/global-themes/digital-disruption/whats-now-and-next-in-analytics-ai-and-automation
14. 14
AI Financing and M&A Activity
WHY AI MATTERS
https://www.cbinsights.com/blog/artificial-intelligence-startup-funding/
https://www.cbinsights.com/blog/top-acquirers-ai-startups-ma-timeline/
15. 15
The Dark Side of AI
WHY AI MATTERS
http://observer.com/2015/08/stephen-hawking-elon-musk-and-bill-gates-warn-about-artificial-intelligence/
https://blog.openai.com/introducing-openai/
Hawking Gates Musk
17. 17
Autonomous delivery vehicles that will reduce the cost of the “last mile”
OPPORTUNITIES
http://dispatch.ai
https://techcrunch.com/2016/04/06/self-driving-delivery-vehicle-startup-dispatch-raises-2-million-seed-round-led-by-andreessen-horowitz
Sensors and AI techniques help Dispatch ensure its delivery vehicle navigates effectively and acts safely around people
18. 18
A shopping experience tailored to you
OPPORTUNITIES
https://metail.com/technology
http://www.techworld.com/startups/how-uk-startup-metail-is-using-computer-vision-change-retail-industry-3654337
Machine learning algorithms and computer vision help Metail show you what a garment will like on you
19. 19
Robots that will take over tedious tasks from humans
OPPORTUNITIES
https://www.bloomberg.com/news/articles/2016-10-26/secretive-canadian-company-teaches-robots-to-be-more-like-people
https://www.theverge.com/2017/6/1/15703146/kindred-orb-robot-ai-startup-warehouse-automation
At Kindred, humans help machines pick up objects and data is used to learn and make the machine smarter over time.
20. 20
Software that will more accurately predict fraud
OPPORTUNITIES
https://fraugster.com/
https://techcrunch.com/2017/01/16/fraugster/
AI-powered fraud detection technology from
Fraugster learns from each transaction in real-time
and is able to anticipate fraudulent transactions before
they occur
21. 21
New drugs and approaches to combat disease
OPPORTUNITIES
http://www.stratifiedmedical.com/about-us/
http://www.economist.com/news/science-and-technology/21713828-silicon-valley-has-squidgy-worlds-biology-and-disease-its-sights-will/
BenevolentAI enables scientific discovery by generating usable knowledge from vast volumes of information in
scientific papers, patents, clinical trial information and from large data sets
22. 22
“
OPPORTUNITIES
https://hbr.org/2016/11/the-simple-economics-of-machine-intelligence/
Machine intelligence is, in its essence, a prediction technology, so the
economic shift will center around a drop in the cost of prediction….
this matters because prediction is an input to a host of activities
including transportation, agriculture, healthcare, energy
manufacturing, and retail… but we will also use prediction to tackle
other problems for which prediction was not historically an input.
— Ajay Agrawal, Joshua Gans, Avi Goldfarb, University of Toronto
24. 24
Those that focus on a narrow domain…
THE WINNERS
http://cdixon.org/2015/02/01/the-ai-startup-idea-maze/
https://medium.com/startup-grind/building-an-ai-startup-realities-tactics-6e1d18a4f7ab/
Applications of AI/ML require vast amounts of data. Scale is beneficial.
• Start-ups inherently disadvantaged in comparison to the big players (e.g. Google, Facebook)
Large companies generally focussed on AI platforms and applications for consumers.
• Focus on the enterprise
• Focus on developing specific tools
Data required is relative to the breadth of the problem you are trying to solve.
• The wider the domain, the more data required
• A narrow domain allows for less data, less likely intrusion from large tech players
25. 25
…and those that can achieve data network effects
THE WINNERS
http://mattturck.com/the-power-of-data-network-effects/
https://medium.com/@muellerfreitag/10-data-acquisition-strategies-for-startups-47166580ee48/
Even with a very specific focus, gathering data required for applications of AI/MI is challenging
• Crowdsourcing data and creating “data traps” potential solutions
• Alternatively, leverage: manual data gathering, public/private data sets, side business
• Goal: achieve data network effects