2. Two truths and a lie
1. Hoofstep – AI for horses
2. Weedguide – AI for cannabis
3. Nailguide – AI for nails
3. Aarthi Srinivasan
- Product Management atTarget Inc
- 15+ years of combined experience in product management, consulting and
engineering
- MBA- The Wharton School, MS in Computer Science – Stony Brook University,
NY & BE – University of Madras, India
Hello
4. 1. AI Funding Trends and future
2. Startup Investment areas
3. Corporate Investment in AI
4. How to bring it home to your company?
What’s on topic today?
5. 5
AI Winters AI Blossom
Images: Google+, Mustdotravels blog
• Computing scale: CPU, GPU, ASICs
• Datasets and infrastructure to handle big
data
• Amazon, Google, FB, MSFT investing in
platforms
6. AI investment growth is unprecedented
Image: CBinsights
15B+ in AI funding in 2017
9. 9
* - 2012 – 2017 ; Ref: CB Insights,Venture Beat,
AI growth areas in cybersecurity & Automation
Start ups
~$15.4B (2012-2017)
Autotech/
Manufacturing
4.0
Facial
Recognition /
Cybersecurity
2
Expert Automation
& Augmentation
Software (EAAS)*
3
1
Healthcare
Diagnostics
& Applied AI
4
$15.4 B* AI startup investments with $15+ Trillion impact on GDP by 2030
iOT AI on the
edge
5
10. 10* - 2012 – 2017 ; Ref: CB Insights,Venture Beat,
Corporate Focus Areas – Alexa & Ok Google our experts
Investments
~$8.5B (2013-2017)
Platform
Leadership
(DIY)
Voice is the new
text
21
AI Talent
Wars &
Research
4
Business
process
automation
3
11. Understand
our Customer
Problem
(internal or
external)
11
Start Small -
Save Cost /
Automation with
error reduction
Try new
solutions in few
channels to
showcase
results
Expand to
other
channels &
problem sets
Create a product organization with agile practices &
transparency with key objectives
1. “Use Deep learning to automate anything a human can do without thinking much in less than 1 second e.g. Identify cat
pictures or smile / frown
2. Use for Predictive analysis “ – Andrew Ng with caveats
Bring AI home to your company
12. 1. Why now?
2. Start up investment themes
3. Corporate Investments
4. How to bring it home to your company?
Questions: saarthi@gmail.com
Summary
14. AI is not new - Why now?
14
Ref – McKinsey Co, MIT Lex Fridman, HBR, Indeed.com
15. 15
Types of AI
MACHINE
LEARNING
Supervised Unsupervised Reinforcement
DEEP
LEARNING
Convolutional Neural
Network – CNN
Recurrent Neural
Network - RNN
Transfer Learning, General Intelligence
16. AI adoption in 3 to 8 years
• Voice recognition & ease of use
• Platforms to prep Training data &
Analyze
• Applied AI in Finance, Security,
Education
• Industry 4.0 (Manufacturing /
Weaponry…)
• Autonomous vehicles (level 3)
• Vision & Emotion precision: E-
commerce (Electronic malls),
Intelligent individual
• Healthcare Disease diagnosis
• AI using Blockchain for iOT edge data
• Distant Doctor / Drug development
• Disease free longevity (cancer
cure)
• Space exploration
• Environment protection (if we live
on Earth)
3+ 8+
17. Sample product purchase evolution
17
Door to door sales or
In-Store
Online & Mobile
platform
Crowd sourced
Marketplace (Amazon,
Ebay, Etsy, Nextdoor,
Facebook)
Predictive Marketplaces
Predictive
Voice
Alexa
Google
Home
iOT
Dash
(Human in the
loop)
Vending
Machines
Robots
Inventory
Robots
Pepper
Robot
24. Identify data needs & create a service mentality
Data Hygienists Clean incoming data for accuracy
E.g. Calendar days vs. working days to count # of days
Data Explorers Sift through data to discover the data we actually need
E.g. Training data
Solution Architects Organize the explored data for analysis & querying
Data Scientists Model the organized data for predictive analytics
Experience
Experts
Turn the models into experiences that get results
E.g. e-mail, Interactive
Reference – HBR
24
25. Unified experience for guests
25
Customer Backbone triggers User Experience with Real-time Context
Customer Lifestage & Shopping Journey
Product
s
Promotions Supply Demand
Prediction
scoring
Context Security
Channel
optimized
Operations
AI / ML Engine (Algorithms, Operations, Network optimizations)
Customer Experience Backend Optimizations
Infrastructure & Data warehouse Management
Segmentation A/B TestingIdentification
Privacy &
Security
Data as a
service
Content & Data Mgmt
User Facing
Tech Backend
P
r
o
d
u
c
t
M
a
n
a
g
e
r
s
Unified experiences through channels such as E-mail, Apps, iOT, Google Home/Alexa, Website, wearables, Stores, at Home
Marketing & Service Channels
9
26. Sources
Aarthi Srinivasan
26
McKinsey,com AI discussion paper
100 AI companies
Business Intelligence Analytics and Trends
When surveillance meets AI
Strategies for Applied AI
https://seekingalpha.com/article/4063499-investing-artificial-intelligence-economic-growth-stock-picking
https://www.cbinsights.com/blog/artificial-intelligence-startup-funding/
https://www.cbinsights.com/blog/cybersecurity-ai-startups-threat-trends/
https://venturebeat.com/2017/06/09/ais-37-billion-market-is-creating-new-industries/
http://www.nanalyze.com/2017/04/6-ai-cybersecurity-startups/
https://www.cbinsights.com/blog/artificial-intelligence-healthcare-startups-investors/
https://www.cbinsights.com/blog/artificial-intelligence-startups-healthcare/
https://www.wired.com/2017/05/mapped-top-263-companies-racing-toward-autonomous-cars/
http://www.businessinsider.com/the-companies-most-likely-to-get-driverless-cars-on-the-road-first-2017-4/#2-
general-motors-17
https://www.techemergence.com/examples-of-artificial-intelligence-in-education/
https://venturebeat.com/2017/03/13/5-tech-firms-racing-to-invest-in-ai-startups/
https://www.cbinsights.com/blog/top-acquirers-ai-startups-ma-timeline/
http://www.cms-connected.com/News-Archive/April-2017/Google-Apple-Facebook-Intel-Microsoft-Salesforce-Twitter-
Battle-AI-Supremacy
https://www.wired.com/insights/2014/08/the-new-eyes-of-surveillance-artificial-intelligence-and-humanizing-
technology/
https://www.movidius.com/news/intel-movidius-helps-bring-artificial-intelligence-to-video-surveillance-cameras
https://www.forbes.com/sites/steveculp/2017/02/15/artificial-intelligence-is-becoming-a-major-disruptive-force-in-
banks-finance-departments/#52bc03a34f62
https://www.forbes.com/sites/gilpress/2017/01/23/top-10-hot-artificial-intelligence-ai-technologies/#58bb8bf19287
https://www.cbinsights.com/research/top-acquirers-ai-startups-ma-timeline/
http://Cbinsights.com 2018 report
Univeristy courses – Stanford / MIT / Coursera
Editor's Notes
Let us start with 2 truths and a Lie. Which one is a lie ?
Sweden’s Hoofstep raised VC money to bring deep learning-based behavioral analysis to horses.
Are you vegan, gluten-free, or allergic to soy? New York’s Prose wants to use AI for made-to-order hair products. It raised $7.57M from well-known VCs including Forerunner Ventures, Lerer Hippeau Ventures, and Maveron.
From brewing beer to tending to cannabis buds, machine learning is doing it all.
AI is also coming to cannabis tech. DeepGreen uses computer vision to identify the gender and health profile of cannabis plants. Weedguide raised $1.7M to use AI for personalized weed recommendations.
I am Aarthi and I enjoy building dynamic teams to launch interesting products. At Target, I am focused on creating a personalized experience for key customer segments across channels and creating a backbone for the customer journey.
Prior to that I was with Walmart Labs leading their personalization strategy & platforms. Earlier I was at Financial Engines where my teams launched the award winning Social Security and Income Planning solutions that eventually resulted in their CEO being invited to the white house. I was also a product manager at Intuit with a focus on behavioral science for customer conversion initiatives and has worked in enterprise applications in my earlier years.
On the academic front, I hold an MBA from Wharton, MS in Computer Science from Stony Brook University and BS from Madras University.
I enjoy motivating teams to combine voice of the customer, data analytics and lean testing to manage a portfolio of products.
Today in the 25 minutes we have, we will explore some funding trends and investment areas in AI.
This talk is successful if you can start visualizing which areas you can start to explore for AI development in your organization
AI is not new but why is it such a hot trend. It is because the compute capability has increased.
Berkeley introduced Spark language and Andrew Ng started using GPUs for computation in 2009. Now we have many more tools to handle large data sets.
Algorithms have become more modern: Backprop, CNN, LSTM (RNN)
Big firms are investing in hardwarde and software Infrastructure to build the world’s leading machine learning AI platform
Chinese company Cambricon is pledging to make one billion processing units in the next three years and is developing chips specifically for deep learning.
It is amazing to be part of the journey where research intersects with reality.
AI funding is growing at an unprecendented rate compared to the overall VC funding which declined at 4% in 2017.
AI funding was about 8 B in startup funding around summer of 2017 and it grew to 15.4 B in funding in 2017 .
In 2016 US was leading all the investments in AI with Silicon Valley leading investments. Just in a span of a year, China has invested heavily in facial recognition technology and hardware in this space to own 48% of the share.
What does this graph depict?
This depicts the January month every year and the number of employees in thousands in manufacturing industry. The point here is you will not have the same level of manufacturing jobs that you had in 2008 due to automation with robotics and it will require a skillset change. Additionally the bereau also predicts the outlook for quality control inspectors and assembler and fabricators, to be negative due to the impact of automation.
Amazon already uses over 100,000 robots in various warehouses, but at the same time is creating thousands of new jobs for humans in its new fulfillment centers.
With 15 Trillion impact by 2030 and 15 B in Investments (9B in startups and 6 – 7B in large companies), we can determine how the funding is distributed.
In the startup space we see the following:
Transportation disruption
Areas here include Services, Safety/Security, In-car intelligence, Autonomy, Infrestructure for connected cars & fleet management, Intelligent manufacturing and Onboard sensors
Simultaneously we will see Manufacturing 4.0 with robots e.g. shoes and Apparel
Computer operated Robotics – Cobots for the manufacturing floor.
Recently, Chinese T-shirt manufacturer Tianyuan Garments Company signed a Memorandum of Understanding (MoU) with the Arkansas government to employ 400 workers at $14/hr at its new garment factory in Arkansas. Operations were scheduled to begin by the end of 2017. Tianyuan’s factory in Little Rock, Arkansas, will use sewing robots developed by Georgia-based startup SoftWear Automation to manufacture apparel for Adidas.
Amazon already uses over 100,000 robots in various warehouses, but at the same time is creating thousands of new jobs for humans in its new fulfillment centers.
In Southern India Chennai there is a restaurant where a robot is bringing in dishes to the customers and taking back dirty dishes. =========
Agro AI automation to care for plants which need water or fertilizers – targeted based on image recognition robotics.
AR/VR to design the manufacturing units
Factory floor person exoskeleton to prevent repetitive injuries
Vision detection to prevent defects
Viant / Everledger / Provenance – Blockchain sourcing
Blockchain: Walmart has been working with IBM since 2016, and said that blockchain technology helped reduce the time required to track mango shipments from 7 days to 2.2 seconds.
The second big theme is about Facial recognition and cybersecurity
A total of 134 startups have raised $3.65B in equity funding in the last 5 years. About 34 of them raised equity for the first time last year to compete in a market still dominated by larger companies like Cybereason, CrowdStrike, Cylance, and Tanium — each with $900M+ valuations.
China’s focus area is on the vision chips for deep learning which will be used with the other focus area namely facial recognition for surveillance. In 2017, around 55 cities in China were part of a plan called Xue Liang or “sharp eyes.” Footage from surveillance cameras in public and private properties will be processed centrally to monitor people and events. Media reports suggest that this may eventually power China’s Social Credit System: a metric to gauge the “trustworthiness” of its citizens.
Startup Megvii already has access to 1.3 billion face data records on Chinese citizens and is backed by Chinese insurance companies (Sunshine Insurance Group), government entities (Russia-China investment group), and corporate giants (Foxconn, Ant Financial).
A growing wave of AI-infused Expert Automation&Augmentation Software (EAAS, pronounced /ēz/) platforms will usher in a new era of AI-assisted or AI-enhanced productivity.This AI-enhanced productivity is threatening jobs at the more clerical end of the white-collar spectrum.
Automation in multiple areas using NLP like Legal search, HR management – Recruting search, Financial management – fund fees, Marketing – email targeting etc.
Andrew Ng suggests that any task that can be done by the human in less than a second can potentially be automated.
Automating healthcare repeatable tasks
Key areas in healthcare are: disease diagnosos – 41%, predictive analytics – 31%, remote monitoring and data analysis – 14%, drug development -12%, prescription analysis – 1% and clinical trials knowledgebase – 1%
Machine learning algorithms can compare a medical Healthcare diagnostics and applied AI are other userful areas. The current wave of applications are all geared to assist radiologists and physicians, as opposed to serving as a final verdict on diagnosis
Compare scans or DNA samples with millions of other patients suggesting diagnosos. Picking up on nuances that a human eye may miss. It can do in seconds what a human would take hours to complete.
Consumer-focused AI monitoring tools like SkinVision — which uses computer vision to monitor suspicious skin lesions — are already in use. But a new wave of healthcare AI applications will institutionalize ML capabilities in hospitals and clinics.
This month AstraZeneca announced a partnership with Alibaba subsidiary Ali Health to develop applications including AI-assisted screening and diagnostics in China.
One of the first FDA approvals was for startup Arterys. Its cloud computing platform was approved for analyzing cardiac images, reportedly based on a series of tests for accuracy and speed of diagnosis. It is now applying for FDA approval for AI in oncology.
Another startup, MedyMatch, is using deep learning to detect intracranial hemorrhage from CT scans. The FDA recently gave it a “breakthrough device designation” to expedite the process of bringing the product to market.
The biggest bone of contention in a high-stakes industry like healthcare is who takes responsibility for misdiagnosis by AI software.
With blockchain and AI, the iOT applications can grow fast. The data will be collected at the edge, validated by the blockchain for accuracy (if latency is not compromised) and computation will be done at the edge. The blockchain will provide the trustworthy backup of data but the AI computation can be done at the edge without the latency to connect to your cloud platform. For example, response in your tesla is immediate or nest / ring response to a package thief. Need a response without internet connectivity.
In the corporate sector we have approximately 9 BB in investments and acquisitions. Common themes here are:
Nearly 120 startups have exited out of which 115 were acquisitions.AI startup acquisitions are up 44%. On average the big 5 Google, Apple, Facebook and Amazon/MSFT have been acquiring 1 company per month each. Intel, salesforce, meltwater outside insight media analytics company (media intelligence to predict what is going to happen or social analytics) and twitter has acquired 5 or 6 companies each as well.
Who becomes the leader in DIY platforms like Google Cloud AutoML, Amazon AI and Microsoft is also in the mix. Still too complex but new students are embracing the data science field.
In spring 2016, UC Berkeley’s first Foundations of Data Science course attracted around 300 students. This semester, nearly 1,000 have enrolled — and university officials are working to create a data science undergraduate major, the first new major for the College of Letters and Science in at least 16 years.
Voice is the new text and we already interact with Alexa and Google using voice. Now chatbots and voice components will be the new way for web interaction
Two years ago, Google acquired AI company DeepMind Technologies for over $500 million, which resulted in 5 percent savings in power usage efficiency and a 40 percent reduction in cooling costs.
DeepMind’s specific focus on improvements in general AI research has helped Google apply the research to its business process to improve its AI capabilities
Kaggle is a crowd sourced platform for solving your AI problems
AI talent wars and at least 150,000 AI jobs as predicted by McKinsey research and Gartner shortages in hiring causing hiring price wars.
Gartner Says More Than 40 Percent of Data Science Tasks Will Be Automated by 2020
Approximately 168K (indeed.com Data Science, ML, AI), By 2018 alone we could be short of about 140K to 190K people with deep analytical skills according to McKinsey Global Institute
Enter Citizens data scientist – Sharron Terry
Different data roles such as Data hygenists, Explores, Solution architects, Data scientists and experience experts
Big data initiatives fail because the internal customers don’t have confidence in the team and don’t trust the models. Trust starts with transparency
Align incentives of the Data team with business teams. Are your models being used by business?
WORRY LESS ABOUT CRUNCHING IT BUT MORE FOCUSSED ON SERVING YOUR MODEL.
Large data set & processing tools
Modern algorithms: Backprop, CNN, LSTM
Infrastructure / Software
Uses training data and feedback from humans to learn the relationship of given inputs to a given output
An algorithm explores input data without being given an explicit output variable
Algorithm learns a task simply by trying to maximize rewards it receives for its actions
Supervisd: Uses training set to make a prediction.
E.g. Model predicts diamond prices based on past prices.
Unsupervised: Provide data without suggesting anything so computer can identify patterns or groupings.
E.g. Customer segmentation, DNA groupings.
Supervised - regression
Using the features provided in the training set make a prediction. Fit a curve using the data provided.
E.g. Price of diamond = X*Cut + Y*Clarity + Z*Size + other features…
GLOBAL FUNDING TOTAL SURGES TO POST-2000 RECORD, PROPELLED BY ASIA
Total annual global funding increased nearly 50% in 2017, as over $164B was invested across 11,042 deals. Deal activity was up by 11%, with both deal and dollar figures representing annual highs – CBInsights & PWC
GLOBAL FUNDING TOTAL SURGES TO POST-2000 RECORD, PROPELLED BY ASIA
Total annual global funding increased nearly 50% in 2017, as over $164B was invested across 11,042 deals. Deal activity was up by 11%, with both deal and dollar figures representing annual highs – CBInsights & PWC