MODERNIZATION OF DIGITAL ENTERPRISES AI AT THE CORE...Apr 1, 2009: An April Fool’s Day joke Nov 5,...
Transcript of MODERNIZATION OF DIGITAL ENTERPRISES AI AT THE CORE...Apr 1, 2009: An April Fool’s Day joke Nov 5,...
MODERNIZATION OF DIGITAL ENTERPRISES
AI AT THE CORE
From Models to Outcomes
Hardik Tiwari, Prateek Das
Mathematicians and Scientists envisioned
the possibilities of predicting outcomes
The world started imagining what if machines are
smarter
Movies, Books, Art
Humankind has always been fascinated by the ability of machines to learn
Thomas Bayes
conceptualized Bayes
Theorem in 1763
Alan Turing
Coined the term
“Turing Test” in
1950
Arthur Samuel
Wrote the first
Machine Learning
code in 1952
Smart Reply
Apr 1, 2009: An April Fool’s Day joke
Nov 5, 2015: Launched real product
Feb 1, 2016: >10% of mobile Inbox
replies
And now we live in a present in which humans and intelligent
systems are bound together in a symbiotic autonomy
AI permeates our daily lives — from
search engines to ride-share
schedulers to ever needful digital
personal assistantsReceived a reminder about
Confluence from Google
Checked route on maps
Booked a cab on Uber
Received a location update for Taj
Taking notes on Evernote
Took a selfie for Instagram
AI has reached a stage where intelligent systems have bettered the humans at times
Face Recognition
97.5%
Human AI/ Machine
Lip reading
41.3%
Pneumonia Detection
75.3%
97.7%
57.9%
75.9%
And now AI technologies have become pervasive in every industry
$25BEstimated revenue
from AI products &
services in 2025
~5MPotential jobs to be
impacted in US by
2025
~10,000Global AI start-ups by
2025
Scale
of
dis
ruption
AI complexity
Sentiment
Analysis
Personalized
Financial
Products
Robo-
Advisors
Credit
Scoring
Fraud
Detection
Automated
Trading
BFSI
Portfolio
Management
Loan/
Insurance
underwriting
Chatbots
Risk
Management
Top Brands
Uses automated analysis to help
identify clients best positioned for
follow-on equity offerings.
Added AI enhancements to its
mobile banking app, which will give
users personalized insights into
their finances.
Cognitive RPA
And now AI technologies have become pervasive in every industry
$25BEstimated revenue
from AI products &
services in 2025
~5MPotential jobs to be
impacted in US by
2025
~10,000Global AI start-ups by
2025
Scale
of
dis
ruption
AI complexity
Hospital
Manageme
nt
Health Analytics
& Prediction
Drug Discovery
Patient
Monitoring
Diagnosis
Robo-assisted
Surgery &
Therapy
Healthcare
Virtual Nursing
Assistant
Virtual
Consultation
has developed a portfolio of AI
solutions that help automate and
standardize complex diagnostics to
meet the needs of every patient.
Medtronic uses IBM Watson for its
remote drug delivery and
monitoring solution for diabetic
patients
Top Brands
And now AI technologies have become pervasive in every industry
$25BEstimated revenue
from AI products &
services in 2025
~5MPotential jobs to be
impacted in US by
2025
~10,000Global AI start-ups by
2025
Scale
of
dis
ruption
AI complexity
Productivit
y
Security &
Surveillance
Marketing
Automation
Enterprise
Software
Automated
Frontend
Development
Data Access
Management
Predictive
Maintenance
Top Brands
Einstein built over its CRM learns from
all that data to deliver predictions and
recommendations based on different
unique business processes
Symantec’s Endpoint Protection 14,
a new security solution harnesses
artificial intelligence to protect clients
And now AI technologies have become pervasive in every industry
$25BEstimated revenue
from AI products &
services in 2025
~5MPotential jobs to be
impacted in US by
2025
~10,000Global AI start-ups by
2025
Scale
of
dis
ruption
AI complexity
Content
Recommendation
Personalization
Customer
Analytics
Retail
Product Marketing
Product Placement
Inventory Planning
& Management
Payment &
Services
Logistics &
Delivery
Lead Generation
Theft Tracking
Walmart partners with Bossa Nova,
whose fully autonomous robots
use machine vision to scan
shelves and monitor inventory
Amazon prime customers can now
order through Alexa
Top Brands
And now AI technologies have become pervasive in every industry and are changing the way we drive, transact, buy, work and - Live
$25BEstimated revenue
from AI products &
services in 2025
~5MPotential jobs to be
impacted in US by
2025
~10,000Global AI start-ups by
2025
Scale
of
dis
ruption
AI complexity
Predictive
Vehicle
Maintenance
Autonomous
Cars
Assisted
Driving
Automotive
Traffic
Management
Operations
Predictive
Maintenance
Auto-Insurance
Tesla is an American EV company
which utilizes AI to offer its customer
self-driving features
Gm uses its Cruise Automation
platform to create self-driving
autonomous vehicles
Top Brands
What is Niramai doing?
Major Challenges
More than 2B Women (> 25 Years) need breast
cancer screening; less than 200M getting
screened every year
Also Enabling
large scale
socio-economic
impact
While these success stories are encouraging, many ML initiatives across global enterprises are not scaling fast enough
Not understanding regulatory aspect
No data collecting/sharing standards
Piloting cool use cases
Still working on workflow automation
Trying to forecast the impossible
Oil and
Gas
Product DevelopmentHyper-Agile | Orchestration
EcosystemHyper-Collaborative
TalentGlobal | Reskilled |Multi Disciplinary
Leadership
Priorities
Enablers
Capabilities
Vision
CustomerNewer Expectations
New Game, New Rules
The DNA of AI Organization is different
Customers now have different expectations from the experience from AI products and services
Customers
Immediate,
responsive serviceConsistency No UI is the new UIPersonalization
Build. Release. Feedback. Iterate . Scale vs Build. Iterate. Release. Scale
Autopilot
Tay.ai and Zo.ai
Enterprises need to release models early, gather
feedback and iterate to improve the productProducts with limitations released early to
gather feedback data
Apple Maps
Accura
cy o
f M
odel
Time taken
80%
95%
Iterate
Launch
Products
Sigmoid : Your buddy at the Confluence
- Built over Weekend DRAUP Hack
- Hopefully, at 80%
www.login.draup.com/sigmoid/
Products
The machine learning product stack is different
Data storageInfrastructure as a
serviceHigh density computing
Data collection & injection Data preparation & binding
Machine Learning Frameworks and
Algorithm Libraries
Machine learning APIs and Advanced
analytics platforms
Messaging Speech VisionNo UI
AI Platform &
Framework
Infrastructure
Tools
Products
And it is more about Orchestration of Platforms
MessagingNo UI
AI Platform &
Framework
Infrastructure
Tools
Voice
Ecosystem
Medtronic leverages open source infrastructure in multiple areas of its product stack
And it is more about Orchestration of Platforms
`
MessagingNo UI
Tools
Platform
Infrastructure
SugarIQ APP
Ecosystem
Competition landscape has changed, competition is on platforms & creating ecosystems rather than companies or customers
Caffe
IBM System
ML
TorchIntel
Trusted Analytic
s Platfor
m
MLLib
CNTK
Bitkit
Deep Learnin
g
H2O
Mahout
Open Cogniti
on Project
Tensor Flow
Spark
Microsoft Computational Network
Toolkit
Facebook FAIR for
torch
Open source contribution from large technology companies
Baidu’s –Warp CTC
GitHUB Stars Repository
103ktensorflow/tensorflow
30.7k fchollet/keras
24.5k bvlc/caffe
16.5k pytorch
14.6k microsoft/cntk
9.1k deeplearning4j
8.3k theano
8.2k caffe2/caffe2
8.2k tflearn/tflearn
7.9k torch
6.6k deepmind/sonnet
Ecosystem
2018 2019 2020 2021 2022 2023 2024 2025
AI Demand AI supply
~100
K
~2.1M
Glo
ba
l J
ob
Op
en
ing
sTech Mafias Own
35% of the AI Talent
There is a war for AI Talent that only Tech Mafias seem to be winning
~44% of the AI talent in
US
~1M
~60K
~1M
Talent
One of the major factors limiting scale is the
inability to acquire and retain the right ML Talent.
This talent is concentrated in a few key locations
and with few large tech companies
~250KInstalled Big Data & AI talent in
G500 companies
92Kinstalled Machine Learning talent in
G500 companies
32%of the 92K employees are working
for
Tech Giants
400KThere is a demand for 400K
Machine Learning developers by
the enterprises and start-ups
10,400
Seattle Area
24,000
Bay Area
2400
Boston
3600
New York
Atlanta
1800
Israel
4600
UK
2200
France
1000
Spain
950
Sao Paulo
3700
Germany
3100
Bangalore
3100
Beijing
1100
Tokyo
450
Singapore
1100
Hyderabad
950
Netherlands
4300
Shanghai1700
Talent
Mumbai
Delhi
Atlanta
Los angels
Chennai
Pittsburgh
Pune
Paris
Cambridge Klon Amsterdam
PhiladelphiaSan Jose
Houston Tampa
Lyon
Orlando
Denver
Shenyang
Singapore
Hyderabad
Detroit
Green
BayMinneapolis
Hong Kong
Guangzhou
Nanjing
Shenzhen
Gainesville
Guatemala
RecifeBogota
Campinas
SantiagoSau Paulo
Lima, peru
Lagos
Durban
Accra
Nairobi
Morocco
Colombo
Jakarta
Brisbane
Sydney
Melbourne
Adelaide
Perth
Seoul
KawasakiChongqing
Chengdu
ChangchunJilin
Buenos Aires
Ho-chi-minhVizag
Surat
Ahmedabad
Dallas
San Diego
Phoenix
Coimbatore
Stockholm
Chandigarh
Bucharest
Cluj
Gdansk
Kolkata
Cairo
Dubai Jaipur
Our “Talent Simulation” predicts diversification of the
available AI Talent – Driven by democratization of AI
education, infra investments, and maturing ecosystems
2018 20302020 2022 2024 20282026
San
Francisco
Seattle
New York
Boston
London
Munich
Tel Aviv
Tokyo
Beijing
ShanghaiBangalore
Talent
130+Talent Hotbeds
~20% Of AI Talent is employed across
tier-2 locations in 2018
37Countries will be home to 1M
Machine learning developers by
2030
And India with its ecosystem and aspirations
Avail
ab
le A
I ta
len
t
2018 2023(E)
~5K
~45K
18.25%
9.99%
25.88%
12.47%
6.50%
Visionary
Methodical
Rational
Persuasive
Imaginative
21.36%
21.91%
6.87%
11.50%
12.70%
Visionary
Challenge driven
Rational
Persuasive
Imaginative
The Leadership traits in AI first organizations are drastically different
Other enterprises AI-driven enterprises
Leadership
Leaders for an AI Future Present
Experiment and Iterate
Collaborate and Nurture Ecosystems
Make Data-Driven Bets
What we have been busy doing : Bringing AI to Enterprise Decision Making
An Enterprise Decision Science
PlatformEmpowering Leaders with Intelligence and
Transactable insights about
- Customers
- Talent
- Ecosystem
- Peers
What are the major investment themes at Ford?
Understand
Customers’ /
Peers’ Priorities
What are the technology buying centers in General
Motors? Evaluate Customers’
buying Hotspots
We, at Zinnov, have started doing our part…
Chennai
Nemili (Tamil Nadu, India)
Zinnov has set up a data
tagging center in the Tier-3
city of Nemili
It has a population of 20K
and is located about 80
miles from Chennai
We have started taking
action…
Nemili
Source: Zinnov’s DRAUP data tagging center