SlideShare a Scribd company logo
ML in finance—the promise and the peril
Charles Elkan
Goldman Sachs
H2O World
October 22, 2019
Deep learning (DL)
 DL has enabled a Kuhnian wave of
scientific progress.
 Especially in vision and in combined
language/vision.
 DL yields higher accuracy inmany
applications, such as forecasting.
Image source: http://www.cs.toronto.edu/~mren/imageqa/
Why is the world excited about AI nowadays?
3
What is deep learning?
4
DL is the current generation of neuralnetworks
Neural networks were invented in the
1950s, inspired (loosely!) by
neuroscience.
Great research advances in the 1980s,
then again in the 2010s.
Current deep learning exploitsmodern
computational power.
Traditional statistical methods are too
simple for massive data, especiallyfor
text, image, and video data.
Image source: https://www.ais.uni-bonn.de/deep_learning/
Now and foreseeably, AI methods cannot achieve deep understanding.
6
What is understanding?
What is shallow understanding?
What is deep understanding?
Operationally, understanding is the ability to answer questions
of many different types correctly.
The ability to answer a limited range of questions thatare
similar to each other.
Understanding that is multilayered, and that involves broad
context, and that uses broad knowledge.
Etymology: The Latin verb intelligere means to understand, with the root meaning “to selectbetween.”
What can deep learning not do?
Is the answer really three? Maybe it isthree
and a half? Or three and twopieces?
Deep understanding combines language and
vision and knowledge and personal experience.
In this example, deep understanding requires
reasoning about occlusion, about gravity, about
social conventions, and more.
Deep understanding
7
Most applications remain brittle
Examples: Face recognition, street sign recognition, question answering:
9
Q: Alexa, how high is MountEverest?
Q: Alexa, where is it located?
A: “ … 29,029 feet …”
A: “The address … is 318 3rd Ave ...”
AI systems need excessive quantities of training examples
Consider the regional meaning of wicked as in “It’s wicked cold outside.”A human needs
only one example, and some thinking, to know when and when not to use thisidiom.
Consequences of shallow understanding
11
How can we make healthcare visits more convenient for patients,
and more efficient for doctors?
1) The patient describes symptoms in freetext.
2) The app uses ML to select follow-ups from a setof standard
questions; the patient answers these.
3) The physician reads the answers, meets the patient, asks formore
information as needed.
4) The physician makes decisions: medicine, advice,referrals.
Only the physician needs deep understanding!
Example: Convenience and productivity in medicine
Source: https://www.geekwire.com/2018/98point6-launches-virtual-clinic-app-solve-americas-primary-care-crisis-ai/
In general, ML is empirical, while traditional quantitative methods aredeductive.
• Monte Carlo methods are randomized, but stilldeductive.
• Combined methods have an empirical part called“calibration.”
ML is useful for tasks with less opportunity fordeduction.
Note: Any method can be a self-fulfillingprophecy!
About ML specifically in finance
14
ML expands what is feasible:
• Taking advantage of new types of data such as text and video.
• Massive models, sometimes with millions of trainablecoefficients.
• Answering new types of questions, and/or maximizing customizedobjectives.
Example: Estimating a market price for a real estateasset:
• It is empirical how different locations, roads, lot sizes, room types, style, etc. influence prices.
• We can estimate the probability distribution of prices--which is not a standard distribution.
• We can use photographs and news articles asinput.
15
ML versus other quantitative methods
Someone has to be a product manager. This person must (ML-specific tasksin blue):
1. Quantify the size of the businessopportunity.
2. Understand exactly what will be useful for business stakeholders, especially customers.
3. Figure out exactly what metrics should be maximized.
4. Understand which data is useful, find this data, and evaluate itsquality.
5. Design how the new system fits into existingsystems.
6. Quantify latency, volume, and other system needs.
7. Help design the user interface.
8. Design guardrails and monitoring.
22
How to drive a project that uses ML?
Example: Consider probability of default (PD and loss given default (LGD). Loans that may default are
restructured, and not recorded as defaults. So are we missing data about true economic losses?
Understanding the true business process
23
24
1. Lack of consensus on a relevant quantitativeobjective.
2. Too few decisions to be made, each one too important. Example: Venture capitalinvestments.
3. Outputs are not accurate enough to be useful. Example: Understanding legallanguage.
4. Real-time data and/or historical data are different, and/or notavailable.
5. The model is not explainable enough for stakeholders to feelcomfortable.
6. The ML method is held to a higher standardthan
humans, or than the current businessprocess.
Risks for an ML application
Do use the latest AI methods to discover patterns in data
Analyze multiple types of data simultaneously
Expect super-human performance in some ways, but not in deep understanding
Look for systems that use the complementary strengths of software and ofhumans
Do not expect genuine artificial intelligence in the foreseeablefuture
But don’t be confident that genuine AI isimpossible
We can’t predict when, but there will be more breakthroughs in research in the future
Speculating about super-intelligence is like speculating about life on otherplanets
32
The promise and the peril
Thank you
Questions welcome!
35

More Related Content

What's hot

Design considerations for machine learning system
Design considerations for machine learning systemDesign considerations for machine learning system
Design considerations for machine learning system
Akemi Tazaki
 
Basics of Soft Computing
Basics of Soft  Computing Basics of Soft  Computing
Basics of Soft Computing
Sangeetha Rajesh
 
machine learning
machine learningmachine learning
machine learning
RaheemUnnisa1
 
Kiran computer
Kiran computerKiran computer
Kiran computer
Kiran Gohil
 
Language First Protocol from QSi
Language First Protocol from QSiLanguage First Protocol from QSi
Language First Protocol from QSi
John O'Gorman
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
AnkanSinha5
 
Evry Digital Advantage IBM keynote on future of retail
Evry Digital Advantage   IBM keynote on future of retailEvry Digital Advantage   IBM keynote on future of retail
Evry Digital Advantage IBM keynote on future of retail
Brandon Jones
 
Chapter1 presentation week1
Chapter1 presentation week1Chapter1 presentation week1
Chapter1 presentation week1Assaf Arief
 
Introduction to Soft Computing
Introduction to Soft ComputingIntroduction to Soft Computing
Introduction to Soft Computing
Geethika Ramani Ravinutala
 
A quick peek into the word of AI
A quick peek into the word of AIA quick peek into the word of AI
A quick peek into the word of AI
Subhendu Dey
 
Mettl ppt[4037]
Mettl ppt[4037]Mettl ppt[4037]
Mettl ppt[4037]
HarshdaGhai
 
Human Evaluation: Why do we need it? - Dr. Sheila Castilho
Human Evaluation: Why do we need it? - Dr. Sheila CastilhoHuman Evaluation: Why do we need it? - Dr. Sheila Castilho
Human Evaluation: Why do we need it? - Dr. Sheila Castilho
Sebastian Ruder
 
Soft computing abstracts
Soft computing abstractsSoft computing abstracts
Soft computing abstractsabctry
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
mailmerk
 
The concept of information
 The concept of information The concept of information
The concept of information
Himani Gupta
 
Thought-powered typing
Thought-powered typing Thought-powered typing
Thought-powered typing
SAishwaryaDinesh
 

What's hot (17)

Design considerations for machine learning system
Design considerations for machine learning systemDesign considerations for machine learning system
Design considerations for machine learning system
 
NLP Ecosystem
NLP EcosystemNLP Ecosystem
NLP Ecosystem
 
Basics of Soft Computing
Basics of Soft  Computing Basics of Soft  Computing
Basics of Soft Computing
 
machine learning
machine learningmachine learning
machine learning
 
Kiran computer
Kiran computerKiran computer
Kiran computer
 
Language First Protocol from QSi
Language First Protocol from QSiLanguage First Protocol from QSi
Language First Protocol from QSi
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
Evry Digital Advantage IBM keynote on future of retail
Evry Digital Advantage   IBM keynote on future of retailEvry Digital Advantage   IBM keynote on future of retail
Evry Digital Advantage IBM keynote on future of retail
 
Chapter1 presentation week1
Chapter1 presentation week1Chapter1 presentation week1
Chapter1 presentation week1
 
Introduction to Soft Computing
Introduction to Soft ComputingIntroduction to Soft Computing
Introduction to Soft Computing
 
A quick peek into the word of AI
A quick peek into the word of AIA quick peek into the word of AI
A quick peek into the word of AI
 
Mettl ppt[4037]
Mettl ppt[4037]Mettl ppt[4037]
Mettl ppt[4037]
 
Human Evaluation: Why do we need it? - Dr. Sheila Castilho
Human Evaluation: Why do we need it? - Dr. Sheila CastilhoHuman Evaluation: Why do we need it? - Dr. Sheila Castilho
Human Evaluation: Why do we need it? - Dr. Sheila Castilho
 
Soft computing abstracts
Soft computing abstractsSoft computing abstracts
Soft computing abstracts
 
Artificial intelligence
Artificial intelligenceArtificial intelligence
Artificial intelligence
 
The concept of information
 The concept of information The concept of information
The concept of information
 
Thought-powered typing
Thought-powered typing Thought-powered typing
Thought-powered typing
 

Similar to Keynote by Charles Elkan, Goldman Sachs - Machine Learning in Finance - The Promise and The Peril - H2O World 2019 NYC

Cognitive Computing.PDF
Cognitive Computing.PDFCognitive Computing.PDF
Cognitive Computing.PDFCharles Quincy
 
Cognitive future part 1
Cognitive future part 1Cognitive future part 1
Cognitive future part 1
Peter Tutty
 
Cognitive future part 1
Cognitive future part 1Cognitive future part 1
Cognitive future part 1
Peter Tutty
 
30 Argumentative Essay Examples In Illustrator Go
30 Argumentative Essay Examples In Illustrator  Go30 Argumentative Essay Examples In Illustrator  Go
30 Argumentative Essay Examples In Illustrator Go
Tanya Williams
 
Challenges in deep learning methods for medical imaging - Pubrica
Challenges in deep learning methods for medical imaging - PubricaChallenges in deep learning methods for medical imaging - Pubrica
Challenges in deep learning methods for medical imaging - Pubrica
Pubrica
 
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
Altimeter, a Prophet Company
 
Intelligence Report: What Works, What Doesn't and How to Fix It
Intelligence Report: What Works, What Doesn't and How to Fix ItIntelligence Report: What Works, What Doesn't and How to Fix It
Intelligence Report: What Works, What Doesn't and How to Fix It
Douglas Bernhardt
 
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
Fwdays
 
SxSW 2015: Key Insights
SxSW 2015: Key InsightsSxSW 2015: Key Insights
SxSW 2015: Key Insights
Digitas Health LifeBrands
 
The Design and Evaluation of DACADE Visual Tool: Theoretical Implications
The Design and Evaluation of DACADE Visual Tool: Theoretical ImplicationsThe Design and Evaluation of DACADE Visual Tool: Theoretical Implications
The Design and Evaluation of DACADE Visual Tool: Theoretical Implications
journalBEEI
 
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
MITAILibrary
 
Unveiling the Power of Machine Learning.docx
Unveiling the Power of Machine Learning.docxUnveiling the Power of Machine Learning.docx
Unveiling the Power of Machine Learning.docx
greendigital
 
Analysis of “what do you do with all this big data” –ted talk by susan etlinger
Analysis of “what do you do with all this big data” –ted talk by susan etlingerAnalysis of “what do you do with all this big data” –ted talk by susan etlinger
Analysis of “what do you do with all this big data” –ted talk by susan etlinger
Darpan Deoghare
 
Mission: Possible! Your cognitive future in government
Mission: Possible! Your cognitive future in governmentMission: Possible! Your cognitive future in government
Mission: Possible! Your cognitive future in government
IBM Government
 
Innovation series 112318
Innovation series 112318Innovation series 112318
Innovation series 112318
Tim Maurer
 
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docxONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
hopeaustin33688
 
Managing information health
Managing information healthManaging information health
Managing information health
1STOUTSOURCE LTD
 
People communicate effectively throughout the organisation
People communicate effectively throughout the organisationPeople communicate effectively throughout the organisation
People communicate effectively throughout the organisation
pludoni GmbH
 
Automated machine learning: the new data science challenge
Automated machine learning: the new data science challengeAutomated machine learning: the new data science challenge
Automated machine learning: the new data science challenge
IJECEIAES
 
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docx
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docxRead 290 Critical Reading as Critical ThinkingOnlineWeek .docx
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docx
catheryncouper
 

Similar to Keynote by Charles Elkan, Goldman Sachs - Machine Learning in Finance - The Promise and The Peril - H2O World 2019 NYC (20)

Cognitive Computing.PDF
Cognitive Computing.PDFCognitive Computing.PDF
Cognitive Computing.PDF
 
Cognitive future part 1
Cognitive future part 1Cognitive future part 1
Cognitive future part 1
 
Cognitive future part 1
Cognitive future part 1Cognitive future part 1
Cognitive future part 1
 
30 Argumentative Essay Examples In Illustrator Go
30 Argumentative Essay Examples In Illustrator  Go30 Argumentative Essay Examples In Illustrator  Go
30 Argumentative Essay Examples In Illustrator Go
 
Challenges in deep learning methods for medical imaging - Pubrica
Challenges in deep learning methods for medical imaging - PubricaChallenges in deep learning methods for medical imaging - Pubrica
Challenges in deep learning methods for medical imaging - Pubrica
 
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
[REPORT PREVIEW] The AI Maturity Playbook: Five Pillars of Enterprise Success
 
Intelligence Report: What Works, What Doesn't and How to Fix It
Intelligence Report: What Works, What Doesn't and How to Fix ItIntelligence Report: What Works, What Doesn't and How to Fix It
Intelligence Report: What Works, What Doesn't and How to Fix It
 
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
Oleksander Krakovetskyi "Explaining a Machine Learning blackbox"
 
SxSW 2015: Key Insights
SxSW 2015: Key InsightsSxSW 2015: Key Insights
SxSW 2015: Key Insights
 
The Design and Evaluation of DACADE Visual Tool: Theoretical Implications
The Design and Evaluation of DACADE Visual Tool: Theoretical ImplicationsThe Design and Evaluation of DACADE Visual Tool: Theoretical Implications
The Design and Evaluation of DACADE Visual Tool: Theoretical Implications
 
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
DALL-E 2 - OpenAI imagery automation first developed by Vishal Coodye in 2021...
 
Unveiling the Power of Machine Learning.docx
Unveiling the Power of Machine Learning.docxUnveiling the Power of Machine Learning.docx
Unveiling the Power of Machine Learning.docx
 
Analysis of “what do you do with all this big data” –ted talk by susan etlinger
Analysis of “what do you do with all this big data” –ted talk by susan etlingerAnalysis of “what do you do with all this big data” –ted talk by susan etlinger
Analysis of “what do you do with all this big data” –ted talk by susan etlinger
 
Mission: Possible! Your cognitive future in government
Mission: Possible! Your cognitive future in governmentMission: Possible! Your cognitive future in government
Mission: Possible! Your cognitive future in government
 
Innovation series 112318
Innovation series 112318Innovation series 112318
Innovation series 112318
 
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docxONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
ONE POINT OF VIEWPaul N. Friga and Richard B. ChapasMA.docx
 
Managing information health
Managing information healthManaging information health
Managing information health
 
People communicate effectively throughout the organisation
People communicate effectively throughout the organisationPeople communicate effectively throughout the organisation
People communicate effectively throughout the organisation
 
Automated machine learning: the new data science challenge
Automated machine learning: the new data science challengeAutomated machine learning: the new data science challenge
Automated machine learning: the new data science challenge
 
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docx
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docxRead 290 Critical Reading as Critical ThinkingOnlineWeek .docx
Read 290 Critical Reading as Critical ThinkingOnlineWeek .docx
 

More from Sri Ambati

GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
Sri Ambati
 
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo DayH2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
Sri Ambati
 
Generative AI Masterclass - Model Risk Management.pptx
Generative AI Masterclass - Model Risk Management.pptxGenerative AI Masterclass - Model Risk Management.pptx
Generative AI Masterclass - Model Risk Management.pptx
Sri Ambati
 
AI and the Future of Software Development: A Sneak Peek
AI and the Future of Software Development: A Sneak Peek AI and the Future of Software Development: A Sneak Peek
AI and the Future of Software Development: A Sneak Peek
Sri Ambati
 
LLMOps: Match report from the top of the 5th
LLMOps: Match report from the top of the 5thLLMOps: Match report from the top of the 5th
LLMOps: Match report from the top of the 5th
Sri Ambati
 
Building, Evaluating, and Optimizing your RAG App for Production
Building, Evaluating, and Optimizing your RAG App for ProductionBuilding, Evaluating, and Optimizing your RAG App for Production
Building, Evaluating, and Optimizing your RAG App for Production
Sri Ambati
 
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
Sri Ambati
 
Risk Management for LLMs
Risk Management for LLMsRisk Management for LLMs
Risk Management for LLMs
Sri Ambati
 
Open-Source AI: Community is the Way
Open-Source AI: Community is the WayOpen-Source AI: Community is the Way
Open-Source AI: Community is the Way
Sri Ambati
 
Building Custom GenAI Apps at H2O
Building Custom GenAI Apps at H2OBuilding Custom GenAI Apps at H2O
Building Custom GenAI Apps at H2O
Sri Ambati
 
Applied Gen AI for the Finance Vertical
Applied Gen AI for the Finance Vertical Applied Gen AI for the Finance Vertical
Applied Gen AI for the Finance Vertical
Sri Ambati
 
Cutting Edge Tricks from LLM Papers
Cutting Edge Tricks from LLM PapersCutting Edge Tricks from LLM Papers
Cutting Edge Tricks from LLM Papers
Sri Ambati
 
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
Sri Ambati
 
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
Sri Ambati
 
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
Sri Ambati
 
LLM Interpretability
LLM Interpretability LLM Interpretability
LLM Interpretability
Sri Ambati
 
Never Reply to an Email Again
Never Reply to an Email AgainNever Reply to an Email Again
Never Reply to an Email Again
Sri Ambati
 
Introducción al Aprendizaje Automatico con H2O-3 (1)
Introducción al Aprendizaje Automatico con H2O-3 (1)Introducción al Aprendizaje Automatico con H2O-3 (1)
Introducción al Aprendizaje Automatico con H2O-3 (1)
Sri Ambati
 
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
Sri Ambati
 
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
Sri Ambati
 

More from Sri Ambati (20)

GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
GenAISummit 2024 May 28 Sri Ambati Keynote: AGI Belongs to The Community in O...
 
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo DayH2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
 
Generative AI Masterclass - Model Risk Management.pptx
Generative AI Masterclass - Model Risk Management.pptxGenerative AI Masterclass - Model Risk Management.pptx
Generative AI Masterclass - Model Risk Management.pptx
 
AI and the Future of Software Development: A Sneak Peek
AI and the Future of Software Development: A Sneak Peek AI and the Future of Software Development: A Sneak Peek
AI and the Future of Software Development: A Sneak Peek
 
LLMOps: Match report from the top of the 5th
LLMOps: Match report from the top of the 5thLLMOps: Match report from the top of the 5th
LLMOps: Match report from the top of the 5th
 
Building, Evaluating, and Optimizing your RAG App for Production
Building, Evaluating, and Optimizing your RAG App for ProductionBuilding, Evaluating, and Optimizing your RAG App for Production
Building, Evaluating, and Optimizing your RAG App for Production
 
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
Building LLM Solutions using Open Source and Closed Source Solutions in Coher...
 
Risk Management for LLMs
Risk Management for LLMsRisk Management for LLMs
Risk Management for LLMs
 
Open-Source AI: Community is the Way
Open-Source AI: Community is the WayOpen-Source AI: Community is the Way
Open-Source AI: Community is the Way
 
Building Custom GenAI Apps at H2O
Building Custom GenAI Apps at H2OBuilding Custom GenAI Apps at H2O
Building Custom GenAI Apps at H2O
 
Applied Gen AI for the Finance Vertical
Applied Gen AI for the Finance Vertical Applied Gen AI for the Finance Vertical
Applied Gen AI for the Finance Vertical
 
Cutting Edge Tricks from LLM Papers
Cutting Edge Tricks from LLM PapersCutting Edge Tricks from LLM Papers
Cutting Edge Tricks from LLM Papers
 
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
Practitioner's Guide to LLMs: Exploring Use Cases and a Glimpse Beyond Curren...
 
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
Open Source h2oGPT with Retrieval Augmented Generation (RAG), Web Search, and...
 
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
KGM Mastering Classification and Regression with LLMs: Insights from Kaggle C...
 
LLM Interpretability
LLM Interpretability LLM Interpretability
LLM Interpretability
 
Never Reply to an Email Again
Never Reply to an Email AgainNever Reply to an Email Again
Never Reply to an Email Again
 
Introducción al Aprendizaje Automatico con H2O-3 (1)
Introducción al Aprendizaje Automatico con H2O-3 (1)Introducción al Aprendizaje Automatico con H2O-3 (1)
Introducción al Aprendizaje Automatico con H2O-3 (1)
 
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
From Rapid Prototypes to an end-to-end Model Deployment: an AI Hedge Fund Use...
 
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
AI Foundations Course Module 1 - Shifting to the Next Step in Your AI Transfo...
 

Recently uploaded

Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
Tobias Schneck
 
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
DanBrown980551
 
GraphRAG is All You need? LLM & Knowledge Graph
GraphRAG is All You need? LLM & Knowledge GraphGraphRAG is All You need? LLM & Knowledge Graph
GraphRAG is All You need? LLM & Knowledge Graph
Guy Korland
 
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered QualitySoftware Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Inflectra
 
Knowledge engineering: from people to machines and back
Knowledge engineering: from people to machines and backKnowledge engineering: from people to machines and back
Knowledge engineering: from people to machines and back
Elena Simperl
 
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
Product School
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
Ana-Maria Mihalceanu
 
Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........
Alison B. Lowndes
 
The Future of Platform Engineering
The Future of Platform EngineeringThe Future of Platform Engineering
The Future of Platform Engineering
Jemma Hussein Allen
 
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
UiPathCommunity
 
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
Ramesh Iyer
 
Connector Corner: Automate dynamic content and events by pushing a button
Connector Corner: Automate dynamic content and events by pushing a buttonConnector Corner: Automate dynamic content and events by pushing a button
Connector Corner: Automate dynamic content and events by pushing a button
DianaGray10
 
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdfFIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance
 
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
Product School
 
Generating a custom Ruby SDK for your web service or Rails API using Smithy
Generating a custom Ruby SDK for your web service or Rails API using SmithyGenerating a custom Ruby SDK for your web service or Rails API using Smithy
Generating a custom Ruby SDK for your web service or Rails API using Smithy
g2nightmarescribd
 
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMsTo Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
Paul Groth
 
Accelerate your Kubernetes clusters with Varnish Caching
Accelerate your Kubernetes clusters with Varnish CachingAccelerate your Kubernetes clusters with Varnish Caching
Accelerate your Kubernetes clusters with Varnish Caching
Thijs Feryn
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
Alan Dix
 
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
Product School
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Albert Hoitingh
 

Recently uploaded (20)

Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
Kubernetes & AI - Beauty and the Beast !?! @KCD Istanbul 2024
 
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
LF Energy Webinar: Electrical Grid Modelling and Simulation Through PowSyBl -...
 
GraphRAG is All You need? LLM & Knowledge Graph
GraphRAG is All You need? LLM & Knowledge GraphGraphRAG is All You need? LLM & Knowledge Graph
GraphRAG is All You need? LLM & Knowledge Graph
 
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered QualitySoftware Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
Software Delivery At the Speed of AI: Inflectra Invests In AI-Powered Quality
 
Knowledge engineering: from people to machines and back
Knowledge engineering: from people to machines and backKnowledge engineering: from people to machines and back
Knowledge engineering: from people to machines and back
 
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
AI for Every Business: Unlocking Your Product's Universal Potential by VP of ...
 
Monitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR EventsMonitoring Java Application Security with JDK Tools and JFR Events
Monitoring Java Application Security with JDK Tools and JFR Events
 
Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........Bits & Pixels using AI for Good.........
Bits & Pixels using AI for Good.........
 
The Future of Platform Engineering
The Future of Platform EngineeringThe Future of Platform Engineering
The Future of Platform Engineering
 
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
Dev Dives: Train smarter, not harder – active learning and UiPath LLMs for do...
 
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
Builder.ai Founder Sachin Dev Duggal's Strategic Approach to Create an Innova...
 
Connector Corner: Automate dynamic content and events by pushing a button
Connector Corner: Automate dynamic content and events by pushing a buttonConnector Corner: Automate dynamic content and events by pushing a button
Connector Corner: Automate dynamic content and events by pushing a button
 
FIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdfFIDO Alliance Osaka Seminar: Overview.pdf
FIDO Alliance Osaka Seminar: Overview.pdf
 
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
De-mystifying Zero to One: Design Informed Techniques for Greenfield Innovati...
 
Generating a custom Ruby SDK for your web service or Rails API using Smithy
Generating a custom Ruby SDK for your web service or Rails API using SmithyGenerating a custom Ruby SDK for your web service or Rails API using Smithy
Generating a custom Ruby SDK for your web service or Rails API using Smithy
 
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMsTo Graph or Not to Graph Knowledge Graph Architectures and LLMs
To Graph or Not to Graph Knowledge Graph Architectures and LLMs
 
Accelerate your Kubernetes clusters with Varnish Caching
Accelerate your Kubernetes clusters with Varnish CachingAccelerate your Kubernetes clusters with Varnish Caching
Accelerate your Kubernetes clusters with Varnish Caching
 
Epistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI supportEpistemic Interaction - tuning interfaces to provide information for AI support
Epistemic Interaction - tuning interfaces to provide information for AI support
 
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
Unsubscribed: Combat Subscription Fatigue With a Membership Mentality by Head...
 
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
Encryption in Microsoft 365 - ExpertsLive Netherlands 2024
 

Keynote by Charles Elkan, Goldman Sachs - Machine Learning in Finance - The Promise and The Peril - H2O World 2019 NYC

  • 1. ML in finance—the promise and the peril Charles Elkan Goldman Sachs H2O World October 22, 2019
  • 2. Deep learning (DL)  DL has enabled a Kuhnian wave of scientific progress.  Especially in vision and in combined language/vision.  DL yields higher accuracy inmany applications, such as forecasting. Image source: http://www.cs.toronto.edu/~mren/imageqa/ Why is the world excited about AI nowadays? 3
  • 3. What is deep learning? 4 DL is the current generation of neuralnetworks Neural networks were invented in the 1950s, inspired (loosely!) by neuroscience. Great research advances in the 1980s, then again in the 2010s. Current deep learning exploitsmodern computational power. Traditional statistical methods are too simple for massive data, especiallyfor text, image, and video data. Image source: https://www.ais.uni-bonn.de/deep_learning/
  • 4. Now and foreseeably, AI methods cannot achieve deep understanding. 6 What is understanding? What is shallow understanding? What is deep understanding? Operationally, understanding is the ability to answer questions of many different types correctly. The ability to answer a limited range of questions thatare similar to each other. Understanding that is multilayered, and that involves broad context, and that uses broad knowledge. Etymology: The Latin verb intelligere means to understand, with the root meaning “to selectbetween.” What can deep learning not do?
  • 5. Is the answer really three? Maybe it isthree and a half? Or three and twopieces? Deep understanding combines language and vision and knowledge and personal experience. In this example, deep understanding requires reasoning about occlusion, about gravity, about social conventions, and more. Deep understanding 7
  • 6. Most applications remain brittle Examples: Face recognition, street sign recognition, question answering: 9 Q: Alexa, how high is MountEverest? Q: Alexa, where is it located? A: “ … 29,029 feet …” A: “The address … is 318 3rd Ave ...” AI systems need excessive quantities of training examples Consider the regional meaning of wicked as in “It’s wicked cold outside.”A human needs only one example, and some thinking, to know when and when not to use thisidiom. Consequences of shallow understanding
  • 7. 11 How can we make healthcare visits more convenient for patients, and more efficient for doctors? 1) The patient describes symptoms in freetext. 2) The app uses ML to select follow-ups from a setof standard questions; the patient answers these. 3) The physician reads the answers, meets the patient, asks formore information as needed. 4) The physician makes decisions: medicine, advice,referrals. Only the physician needs deep understanding! Example: Convenience and productivity in medicine Source: https://www.geekwire.com/2018/98point6-launches-virtual-clinic-app-solve-americas-primary-care-crisis-ai/
  • 8. In general, ML is empirical, while traditional quantitative methods aredeductive. • Monte Carlo methods are randomized, but stilldeductive. • Combined methods have an empirical part called“calibration.” ML is useful for tasks with less opportunity fordeduction. Note: Any method can be a self-fulfillingprophecy! About ML specifically in finance 14
  • 9. ML expands what is feasible: • Taking advantage of new types of data such as text and video. • Massive models, sometimes with millions of trainablecoefficients. • Answering new types of questions, and/or maximizing customizedobjectives. Example: Estimating a market price for a real estateasset: • It is empirical how different locations, roads, lot sizes, room types, style, etc. influence prices. • We can estimate the probability distribution of prices--which is not a standard distribution. • We can use photographs and news articles asinput. 15 ML versus other quantitative methods
  • 10. Someone has to be a product manager. This person must (ML-specific tasksin blue): 1. Quantify the size of the businessopportunity. 2. Understand exactly what will be useful for business stakeholders, especially customers. 3. Figure out exactly what metrics should be maximized. 4. Understand which data is useful, find this data, and evaluate itsquality. 5. Design how the new system fits into existingsystems. 6. Quantify latency, volume, and other system needs. 7. Help design the user interface. 8. Design guardrails and monitoring. 22 How to drive a project that uses ML?
  • 11. Example: Consider probability of default (PD and loss given default (LGD). Loans that may default are restructured, and not recorded as defaults. So are we missing data about true economic losses? Understanding the true business process 23
  • 12. 24 1. Lack of consensus on a relevant quantitativeobjective. 2. Too few decisions to be made, each one too important. Example: Venture capitalinvestments. 3. Outputs are not accurate enough to be useful. Example: Understanding legallanguage. 4. Real-time data and/or historical data are different, and/or notavailable. 5. The model is not explainable enough for stakeholders to feelcomfortable. 6. The ML method is held to a higher standardthan humans, or than the current businessprocess. Risks for an ML application
  • 13. Do use the latest AI methods to discover patterns in data Analyze multiple types of data simultaneously Expect super-human performance in some ways, but not in deep understanding Look for systems that use the complementary strengths of software and ofhumans Do not expect genuine artificial intelligence in the foreseeablefuture But don’t be confident that genuine AI isimpossible We can’t predict when, but there will be more breakthroughs in research in the future Speculating about super-intelligence is like speculating about life on otherplanets 32 The promise and the peril