Download the slides: https://www.hfsresearch.com/webinars/webinar-reality-check-on-enterprise-artificial-intelligence-ai
Watch all HfS webinars: https://www.hfsresearch.com/webinars/
The noise in the market around AI is deafening. Yet, most of this hype is focused on more consumer-facing issues or projects that cannot easily be replicated. To separate the wheat from the chaff HfS’ inaugural Enterprise AI Blueprint has taken stock where organizations really are on their journey toward the OneOffice and how AI is accelerating that journey.
We will present the main findings of the study and discuss the key issues with thought-leaders in the AI space:
- Jesus Mantas, Global Head of Strategy & Offerings, IBM Global Business Services
- Mike Salvino, Managing Director, Carrick Capital Partners, Executive Chairman, Infinia ML
- Phil Fersht, CEO and Chief Analyst, HfS Research
- Tom Reuner, Managing Partner, Business Operations Strategy
26. IBM Services
Enterprise-Grade AI
HfS Webinar
Jesus Mantas | Managing Partner, Cognitive Assets and GBS Ventures Global Head
of Strategy & Offerings, IBM Services
Webinar | April 12, 2018
27. 2727
Trusted, Transparent &
Auditable
• Transparency of training
• Ability to explain decisions
• Auditability of
recommendations
Integrates with work
flows and talent
flows
• Business Purpose
• Secure
• Bidirectional human-
system design
Learns more from less
data
• Limited, not semantically
consistent
• Thousands, not billions
• Industry-context specific
Enterprise-Grade Artificial Intelligence
2
28. 28
Data is the natural resource required for ML or AI to distill any
outcomes
29. 2929
Risk & ComplianceCognitive Care
Knowledge Worker
Enterprise-Grade Artificial Intelligence Use Cases at Scale
HR / Talent
Global Financial Institution
31. The ML Epidemic
Epidemic: a widespread occurrence of a particular undesirable phenomenon
The undesirable phenomenon here is that all of a sudden everyone knows
how to do ML and is an Expert. (The ML Epidemic)
Don’t Fuel the ML Epidemic!
32. ML Epidemic–Point #1: I have a TEAM that knows ML!
Your team that knows how to do Simple Neural Networks…An example is converting English words to
French…These successful simple projects will inspire enterprises to want more out of ML and this is when
your ML team will STRUGGLE. Building Deep Learning Neural Networks (Multi-Layered) is not easy.
33. ML Epidemic–Point #2: The ML projects my team are
performing will scale to achieve true business IMPACT!
The AlphaGo breakthrough was great PR (there is now a documentary on Netflix) but provided little business impact.
Make sure your projects are answering these 3 questions. NO SCIENCE EXPERIMENTS!
1. Does the project solve a top 1, 2, or 3
question that a CEO or Executive wants
answered?
2. Does the project help your company to
reduce costs or increase revenue?
3. Does the project create a unique data
set for your company?
34. ML Epidemic–Point #3: My Data is READY to go!
There is NO ML without Data. Salvino prediction: “Companies will spend as much if not more money dealing
with Data in the next 5 years as they have spent implementing mission critical systems like SAP, etc.”
1. Accessible – Can your ML team actually access your data?
2. Clean – Is your data clean or is there “junk” in fields?
3. Data Sets – Have you created data sets or have you created data “swamps”?
4. Maintain – Do you have a process and team to maintain the data (Data Science Culture)?
5. Utilize – Do you have a strategy to answer question that make business impact?
36. Scarcity of resources is real and expensive. NIPS was a recruiting event this year instead of a research conference. Most
luminaries are not teaching any longer. They are tied up doing work for companies so new resources are not being created to
keep up with demand. Proficiency is developed by doing years of research and most companies don’t have access to labs.
How many ML experts have you met this year that were Cloud or Security experts last year?
How to Get Started–Point #1: EVALUATE your ML talent
1. Advanced degree in relevant quantitative field (statistics, computer science,
applied mathematics, etc.)
2. 7+ years experience in machine learning, data science, data engineering, and/or
computational software development
3. 3+ years development in Python, including libraries such as NumPy, SciPy,
pandas, TensorFlow, etc.
4. Experience with deep learning models, including CNN and RNN architectures
5. Experience working with large datasets, including NoSQL and relational databases
6. Experience with cloud computing
37. Big data, IoT, Analytics, Digital, etc. All good buzz words but it really is not that hard.
Create Data Sets and a Data Science Culture. It is not GLAMOROUS work but it matters!
How to Get Started–Point #2: Create a coherent data strategy across
your data warehouses, lakes, rivers, streams, puddles, swamps…
38. Magicians vs. Aliens. Magicians want to work with other Magicians not folks that view them as Aliens. This is
not inspiring to them. People leave People – they don’t leave Companies so ORGANIZE for success!
How to Get Started–Point #3: CENTRALIZE the ML Function