SlideShare a Scribd company logo
1 of 41
Download to read offline
Manish Gupta
Google Research India
The Transformative
Power of AI and
Open Challenges
Acknowledgment: Work done by several colleagues at Google Research
● AI Trends
○ Deep Learning (dominant paradigm within ML)
○ Foundation Models (new paradigm within DL)
● AI for Synthesis and Creativity
● Inclusive AI
○ Languages
○ Health
○ Agriculture
● Some Challenges and Takeaways
Outline
“cat”
Deep Learning
Modern Reincarnation of Artificial Neural Networks
Collection of simple trainable mathematical units, organized in layers, that work together to solve
complicated tasks
Key Benefit
Learns features from raw, heterogeneous, noisy data
No explicit feature engineering required
What’s New
new network architectures,
new training math, scale
Deep Learning Phase 1: A decade of amazing progress
in what computers can do
Input Output
Pixels:
Audio:
Pixels:
“Hello, how are you?”
“leopard”
“How cold is it outside?”
“Bonjour, comment allez-vous?”
“A cheetah lying on top of a car”
● AI Trends
○ Deep Learning (dominant paradigm within ML)
○ Foundation Models (new paradigm within DL)
● AI for Synthesis and Creativity
● Inclusive AI
○ Languages
○ Health
○ Agriculture
● Some Challenges and Takeaways
Outline
Large, pretrained
models with
self-supervision
Example: BERT
Obama was born in 1961 in Honolulu , Hawaii , two years after the
territory was admitted to the Union as the 50th state . Raised largely in
Hawaii , he also spent one year of his childhood in Washington state
and four years in Indonesia.
v1
v2
v3
v4
.. v11
v12
….v20
… v24
... v33
…..v47
Masks (3, 11, 12,
20, 24, 29, 33)
Encode
Predict masked
words
BERT
Original
words
Masked
words
Obama was ____ in 1961 in Honolulu , Hawaii , ____ ____ after the
territory was admitted to the ___ as the 50th ____ . Raised largely in
____ , he also spent one year of his ____ in Washington state and four
years in Indonesia.
born two years Union state Hawaii childhood

Language models
Chain of Thought
Chain of Thought Prompting Elicits Reasoning in Large Language Models, Jason Wei, Xuezhi Wang, Dale
Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou, https://arxiv.org/abs/2201.11903
Chain of Thought
Chain of Thought Prompting Elicits Reasoning in Large Language Models, Jason Wei, Xuezhi Wang, Dale
Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou, https://arxiv.org/abs/2201.11903
Labeled medical images
How can we utilise unlabelled images?
Large natural datasets
Self and semi supervised learning (SSL)
Unlabeled medical images
Big Transfer (BiT)
Chest X-Ray
interpretation
Breast
cancer
detection
Skin
condition
diagnosis
Gemini
Multimodal
Reasoning
Gemini: A Family of Highly Capable Multimodal Models, by the Gemini Team, arxiv.org/abs/2312.11805
Gemini
Multimodal
Reasoning
● AI Trends
○ Deep Learning (dominant paradigm within ML)
○ Foundation Models (new paradigm within DL)
● AI for Synthesis and Creativity
● Inclusive AI
○ Languages
○ Health
○ Agriculture
● Some Challenges and Takeaways
Outline
Transform and simplify the
creative process with AI
AI image synthesis is the biggest
innovation in image making since
the invention of photography 200
years ago.
- Fredo Durand, MIT
“Abstract surreal painting of
a melting white 1974
Porsche 911 oozing onto a
chess board. Chess pieces
are scattered around. A
melting golden clock shines
brightly in the background.
Soft and eery lighting.”
simple destructive process slowly maps data to noise
[Image credit: Ben Poole]
[Jascha Sohl-Dickstein et al., 2015]
[Yang Song & Stefano Ermon, 2019]
[Jonathan Ho et al., 2020]
Diffusion model is trained to map noise back to data
Diffusion models
[Nanxin Chen et al., 2020]
[Durk Kingma & Tim Salimans et al., 2021]
[Prafulla Dhariwal, Alex Nichol, Aditya Ramesh et al., 2020+]
Google Confidential
Model scaling brings consistent quality wins
A portrait photo of a kangaroo wearing an orange hoodie and blue sunglasses standing on the grass
in front of the Sydney Opera House holding a sign on the chest that says Welcome Friends!

DreamBooth
Image Extrapolation – Dramatic Uncrop
MaskGIT: [Chang et al., CVPR'22]
Agriculture Efficiency of LLMs
Inclusive AI
Information in
Indian Languages
Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
Agriculture Efficiency of LLMs
Inclusive AI
Information in
Indian Languages
Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
English in India Superset of formal English
Variants in query logs User Intent
sift dejar price Swift Dzire price
boys shuj boys shoes
baground piccher background picture
dabal bed farnichar double bed furniture
mehndi dijain mehndi design
IndoPhoneme: Generate phonetic
misspellings influenced by user’s
native language
IndoPhoneme
Generate phonetic misspellings influenced by users native language.
Tamil Speaker
income -> ingam
Hindi Speaker
shoes -> shuj
Corrupted
word
Correct word
Phoneme sequence of the
correct word
Corrupted phoneme
sequence
How do we mine phoneme corruptions?
Synthetic
Noisifiers
Round Trip Transliteration
Noisy Channel Model
Launched as part of Google Search spell correction
Multilingual Language Models
MuRIL: Multilingual Representations for Indian Languages
● BERT for Indian languages. Two versions: Base (236M), Large (550M)
● Largest coverage of Indian languages (17, including English)
● Supports native and Latin transliterated text
+27.8%
Available at
TensorFlow Hub
HuggingFace
arxiv.org/abs/2103.10730
+10.1%
காைல 8 மணிக
் கு அலாரம
்
CREATE_ALARM(DATE_TIME(8 AM))
Multilingual
Semantic Parser
English alarm at 8 am
Urdu ‫اﻻرم‬ ‫ﮐﺎ‬ ‫ﺑﺟﮯ‬ 8 ‫ﺻﺑﺢ‬
Bengali সকাল ৮টায় এলামর্ম
Hindi स
ु बह 8 बज
े अलाम
र्म
Tamil
MuRIL in Google Assistant launched for 8 Indian Languages
Proprietary + Confidential
M rni: Multimodal Representation for India
Morni Vision
Build the best multimodal representation for
100+ Indic languages and accelerate
development of inclusive and equitable Indic
language technologies
Population: 1.4B
Languages: 60 (1M+), 125 (100K+)
Google ASR: 11 languages
go/morni
Proprietary + Confidential
Vaani: Capturing the Speech Landscape in India
A given language varies across
region, collecting data anchored
on language doesn’t capture on
the ground diversity
Vaani approach
Collect image-prompted speech data anchored on
region, rather than language, while ensuring diversity
15.4K
transcribed
773
districts
154K
speech
Open
Source
vaani.iisc.ac.in
PoC: partha@, dineshtewari@
Proprietary + Confidential
LLM Composability
go/lm-composition
PoC: brachit@
Making LLM reasoning abilities available for low-resource languages through composition
Encapsulating org-wide focus areas of model unification, re-use, and modularity.
Emphasising minimal model interventions, data and compute efficiency, and flexibility.
Proprietary + Confidential
Proprietary + Confidential
Agriculture Efficiency of LLMs
Inclusive AI
Information in
Indian Languages
Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
Agricultural Landscape Understanding & Field Event Detection (from Satellite Imagery)
Foundational layer of an agri-stack - use cases like farmer loans, crop insurance, incentives to
avoid stubble burning, carbon modeling to incentivize regenerative agricultural practices
(demo)
Agriculture Efficiency of LLMs
Inclusive AI
Information in
Indian Languages
Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
Rigidity in ML Models
By 2026, more than 80% of enterprises will have used generative artificial intelligence (GenAI)
application programming interfaces (APIs) or models, and/or deployed GenAI-enabled applications
in production environments, up from less than 5% in 2023, according to Gartner, Inc.
….
Analogous to
pre-virtualization era when
workloads would run on a
fixed set of compute
resources (e.g., fixed
number of processors)
Vision: Enable Elasticity of Foundation Models
Accuracy Cost Latency
Gemini right sized model
Without any additional training!
Deliver system level efficiencies and business enablement similar to what
Cloud did for enterprise workloads
Proprietary + Confidential
MatFormer: Matryoshka Transformer
● MatFormer builds upon MRL
● Apply MRL to MLP layer in each transformer block
L-th layer token x
W1
W2
L+1-th layer token x
That is, neurons are nested in each other.
Model-S
Model-XL
Model-M
Model-L
Transformer
Block 1
Transformer
Block 2
Transformer
Block 3
Transformer
Block 4
Transformer
Block 5
Transformer
Block 6
Model-M
Model-S
Mix'n'Match & Routing on MatFormer
● Mix’n’Match: 100s (combinatorial) of static (on-demand) models for all accuracy-compute
● Routing: Token based routing akin to MoE to realize dynamic computation
Matformer: Evaluation
1-shot GPT-evals
● Almost matching accuracy for MatFormer–[XL, L, M, S] models against Baselines
● We get all the intermediate models denotes by ★ for “free”
○ No extra training!
Some Outstanding Challenges
● Robustness
● Safety
● Bias and Fairness
● Explainability
Objectives for AI Applications
1. Be socially beneficial.
2. Avoid creating or reinforcing unfair bias.
3. Be built and tested for safety.
4. Be accountable to people.
5. Incorporate privacy design principles.
6. Uphold high standards of scientific
excellence.
7. Be made available for uses that accord
with these principles.
AI Applications We Will Not Pursue
1. Technologies that cause or are likely to
cause overall harm.
2. Weapons or other technologies whose
principal purpose or implementation is
to cause or directly facilitate injury to
people.
3. Technologies that gather or use
information for surveillance violating
internationally accepted norms.
4. Technologies whose purpose
contravenes widely accepted principles
of international law and human rights.
Google’s AI Principles
https://ai.google/principles/
● AI is here - already impacting every industry and the world at large
● Rise of Foundation Models - models pre-trained on large data,
usually with self-supervision
● Huge opportunity to develop AI models and solutions to bring
benefits to billions of new people - outstanding challenges
Takeaways
These are exciting times!

More Related Content

What's hot

Azure Migration Program Pitch Deck
Azure Migration Program Pitch DeckAzure Migration Program Pitch Deck
Azure Migration Program Pitch DeckNicholas Vossburg
 
Descending from the architect's ivory tower
Descending from the architect's ivory towerDescending from the architect's ivory tower
Descending from the architect's ivory towerValtech UK
 
Andy Jassy Illuminates Amazon Web Services
Andy Jassy Illuminates Amazon Web ServicesAndy Jassy Illuminates Amazon Web Services
Andy Jassy Illuminates Amazon Web ServicesMichael Skok
 
Agile Process Introduction
Agile Process IntroductionAgile Process Introduction
Agile Process IntroductionNguyen Hai
 
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저Amazon Web Services Korea
 
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...Amazon Web Services
 
Cloud Computing
Cloud ComputingCloud Computing
Cloud ComputingOmar Fathy
 
AManaging Kong API Gateway with Terraform
AManaging Kong API Gateway with TerraformAManaging Kong API Gateway with Terraform
AManaging Kong API Gateway with TerraformByungjin Park
 
Building Data Lakes and Analytics on AWS
Building Data Lakes and Analytics on AWSBuilding Data Lakes and Analytics on AWS
Building Data Lakes and Analytics on AWSAmazon Web Services
 
Transform Agile Development With Practical DevOps
Transform Agile Development With Practical DevOpsTransform Agile Development With Practical DevOps
Transform Agile Development With Practical DevOpsGaurav Sharma
 
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018Amazon Web Services Korea
 
Amazon Web Services - Elastic Beanstalk
Amazon Web Services - Elastic BeanstalkAmazon Web Services - Elastic Beanstalk
Amazon Web Services - Elastic BeanstalkAmazon Web Services
 
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...Amazon Web Services Korea
 
Introduction to Google Cloud Platform
Introduction to Google Cloud PlatformIntroduction to Google Cloud Platform
Introduction to Google Cloud Platformdhruv_chaudhari
 

What's hot (20)

Azure Migration Program Pitch Deck
Azure Migration Program Pitch DeckAzure Migration Program Pitch Deck
Azure Migration Program Pitch Deck
 
Descending from the architect's ivory tower
Descending from the architect's ivory towerDescending from the architect's ivory tower
Descending from the architect's ivory tower
 
User Story
User StoryUser Story
User Story
 
Andy Jassy Illuminates Amazon Web Services
Andy Jassy Illuminates Amazon Web ServicesAndy Jassy Illuminates Amazon Web Services
Andy Jassy Illuminates Amazon Web Services
 
Atlassian JIRA
Atlassian JIRAAtlassian JIRA
Atlassian JIRA
 
Agile Process Introduction
Agile Process IntroductionAgile Process Introduction
Agile Process Introduction
 
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저
AWS 클라우드 비용 최적화를 위한 TIP - 임성은 AWS 매니저
 
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...
Build accurate training datasets with Amazon SageMaker Ground Truth - AIM305 ...
 
Google cloud platform
Google cloud platformGoogle cloud platform
Google cloud platform
 
Agile
AgileAgile
Agile
 
Cloud Computing
Cloud ComputingCloud Computing
Cloud Computing
 
AManaging Kong API Gateway with Terraform
AManaging Kong API Gateway with TerraformAManaging Kong API Gateway with Terraform
AManaging Kong API Gateway with Terraform
 
Jira training
Jira trainingJira training
Jira training
 
Building Data Lakes and Analytics on AWS
Building Data Lakes and Analytics on AWSBuilding Data Lakes and Analytics on AWS
Building Data Lakes and Analytics on AWS
 
Transform Agile Development With Practical DevOps
Transform Agile Development With Practical DevOpsTransform Agile Development With Practical DevOps
Transform Agile Development With Practical DevOps
 
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018
천만 사용자를 위한 AWS 클라우드 아키텍처 진화하기::이창수::AWS Summit Seoul 2018
 
Amazon Web Services - Elastic Beanstalk
Amazon Web Services - Elastic BeanstalkAmazon Web Services - Elastic Beanstalk
Amazon Web Services - Elastic Beanstalk
 
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...
마이그레이션과 함께 시작되는 Cloud Financial Management 전략 세우기-곽내인, AWS Cloud Financial Ma...
 
Building Your Cloud Strategy
Building Your Cloud StrategyBuilding Your Cloud Strategy
Building Your Cloud Strategy
 
Introduction to Google Cloud Platform
Introduction to Google Cloud PlatformIntroduction to Google Cloud Platform
Introduction to Google Cloud Platform
 

Similar to "The Transformative Power of AI and Open Challenges" by Dr. Manish Gupta, Google

Landscape of AI/ML in 2023
Landscape of AI/ML in 2023Landscape of AI/ML in 2023
Landscape of AI/ML in 2023HyunJoon Jung
 
building intelligent systems with large scale deep learning
building intelligent systems with large scale deep learningbuilding intelligent systems with large scale deep learning
building intelligent systems with large scale deep learningmustafa sarac
 
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...AI Frontiers
 
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn..."Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...Edge AI and Vision Alliance
 
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...Maryam Farooq
 
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptx
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptxLiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptx
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptxVishnuRajuV
 
Cognitive Assistants - Opportunities and Challenges - slides
Cognitive Assistants - Opportunities and Challenges - slidesCognitive Assistants - Opportunities and Challenges - slides
Cognitive Assistants - Opportunities and Challenges - slidesHamid Motahari
 
Beyond Siri on the iPhone: How could intelligent systems change the way we in...
Beyond Siri on the iPhone: How could intelligent systems change the way we in...Beyond Siri on the iPhone: How could intelligent systems change the way we in...
Beyond Siri on the iPhone: How could intelligent systems change the way we in...Yousif Almas
 
Deep Learning & NLP: Graphs to the Rescue!
Deep Learning & NLP: Graphs to the Rescue!Deep Learning & NLP: Graphs to the Rescue!
Deep Learning & NLP: Graphs to the Rescue!Roelof Pieters
 
Week1- Introduction.pptx
Week1- Introduction.pptxWeek1- Introduction.pptx
Week1- Introduction.pptxfahmi324663
 
Rise of AI through DL
Rise of AI through DLRise of AI through DL
Rise of AI through DLRehan Guha
 
Interpretable Machine Learning
Interpretable Machine LearningInterpretable Machine Learning
Interpretable Machine LearningSri Ambati
 
How can text-mining leverage developments in Deep Learning? Presentation at ...
How can text-mining leverage developments in Deep Learning?  Presentation at ...How can text-mining leverage developments in Deep Learning?  Presentation at ...
How can text-mining leverage developments in Deep Learning? Presentation at ...jcscholtes
 
Bridging the gap between AI and UI - DSI Vienna - full version
Bridging the gap between AI and UI - DSI Vienna - full versionBridging the gap between AI and UI - DSI Vienna - full version
Bridging the gap between AI and UI - DSI Vienna - full versionLiad Magen
 
A-STUDY-ON-SENTIMENT-POLARITY.pdf
A-STUDY-ON-SENTIMENT-POLARITY.pdfA-STUDY-ON-SENTIMENT-POLARITY.pdf
A-STUDY-ON-SENTIMENT-POLARITY.pdfSUDESHNASANI1
 
Deep Learning for AI - Yoshua Bengio, Mila
Deep Learning for AI - Yoshua Bengio, MilaDeep Learning for AI - Yoshua Bengio, Mila
Deep Learning for AI - Yoshua Bengio, MilaLucidworks
 
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...MaRS Discovery District
 
Language Translator.pptx
Language Translator.pptxLanguage Translator.pptx
Language Translator.pptxMRABC9
 

Similar to "The Transformative Power of AI and Open Challenges" by Dr. Manish Gupta, Google (20)

Landscape of AI/ML in 2023
Landscape of AI/ML in 2023Landscape of AI/ML in 2023
Landscape of AI/ML in 2023
 
building intelligent systems with large scale deep learning
building intelligent systems with large scale deep learningbuilding intelligent systems with large scale deep learning
building intelligent systems with large scale deep learning
 
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...
Training at AI Frontiers 2018 - Ni Lao: Weakly Supervised Natural Language Un...
 
Mattingly "Text and Data Mining: Building Data Driven Applications"
Mattingly "Text and Data Mining: Building Data Driven Applications"Mattingly "Text and Data Mining: Building Data Driven Applications"
Mattingly "Text and Data Mining: Building Data Driven Applications"
 
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn..."Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...
"Large-Scale Deep Learning for Building Intelligent Computer Systems," a Keyn...
 
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...
NYAI #27: Cognitive Architecture & Natural Language Processing w/ Dr. Catheri...
 
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptx
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptxLiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptx
LiDeng-BerlinOct2015-ASR-GenDisc-4by3.pptx
 
Cognitive Assistants - Opportunities and Challenges - slides
Cognitive Assistants - Opportunities and Challenges - slidesCognitive Assistants - Opportunities and Challenges - slides
Cognitive Assistants - Opportunities and Challenges - slides
 
Beyond Siri on the iPhone: How could intelligent systems change the way we in...
Beyond Siri on the iPhone: How could intelligent systems change the way we in...Beyond Siri on the iPhone: How could intelligent systems change the way we in...
Beyond Siri on the iPhone: How could intelligent systems change the way we in...
 
Deep Learning & NLP: Graphs to the Rescue!
Deep Learning & NLP: Graphs to the Rescue!Deep Learning & NLP: Graphs to the Rescue!
Deep Learning & NLP: Graphs to the Rescue!
 
Week1- Introduction.pptx
Week1- Introduction.pptxWeek1- Introduction.pptx
Week1- Introduction.pptx
 
Rise of AI through DL
Rise of AI through DLRise of AI through DL
Rise of AI through DL
 
Interpretable Machine Learning
Interpretable Machine LearningInterpretable Machine Learning
Interpretable Machine Learning
 
How can text-mining leverage developments in Deep Learning? Presentation at ...
How can text-mining leverage developments in Deep Learning?  Presentation at ...How can text-mining leverage developments in Deep Learning?  Presentation at ...
How can text-mining leverage developments in Deep Learning? Presentation at ...
 
Bridging the gap between AI and UI - DSI Vienna - full version
Bridging the gap between AI and UI - DSI Vienna - full versionBridging the gap between AI and UI - DSI Vienna - full version
Bridging the gap between AI and UI - DSI Vienna - full version
 
A-STUDY-ON-SENTIMENT-POLARITY.pdf
A-STUDY-ON-SENTIMENT-POLARITY.pdfA-STUDY-ON-SENTIMENT-POLARITY.pdf
A-STUDY-ON-SENTIMENT-POLARITY.pdf
 
Deep Learning for AI - Yoshua Bengio, Mila
Deep Learning for AI - Yoshua Bengio, MilaDeep Learning for AI - Yoshua Bengio, Mila
Deep Learning for AI - Yoshua Bengio, Mila
 
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...
Deep Learning: Changing the Playing Field of Artificial Intelligence - MaRS G...
 
graziani_bias.pdf
graziani_bias.pdfgraziani_bias.pdf
graziani_bias.pdf
 
Language Translator.pptx
Language Translator.pptxLanguage Translator.pptx
Language Translator.pptx
 

More from ISPMAIndia

"Scaling Product Leadership with AI" by Ravi Padaki
"Scaling Product Leadership with AI" by Ravi Padaki"Scaling Product Leadership with AI" by Ravi Padaki
"Scaling Product Leadership with AI" by Ravi PadakiISPMAIndia
 
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee..."Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...ISPMAIndia
 
Building Products That Think- Bhaskaran Srinivasan & Ashish Gupta
Building Products That Think- Bhaskaran Srinivasan & Ashish GuptaBuilding Products That Think- Bhaskaran Srinivasan & Ashish Gupta
Building Products That Think- Bhaskaran Srinivasan & Ashish GuptaISPMAIndia
 
AI Product Management by Abhijit Bendigiri
AI Product Management by Abhijit BendigiriAI Product Management by Abhijit Bendigiri
AI Product Management by Abhijit BendigiriISPMAIndia
 
SPM 2024 – Overview of and benefits of AI in Product Management
SPM 2024 – Overview of and benefits of AI in Product ManagementSPM 2024 – Overview of and benefits of AI in Product Management
SPM 2024 – Overview of and benefits of AI in Product ManagementISPMAIndia
 
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ..."Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...ISPMAIndia
 
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...ISPMAIndia
 
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...ISPMAIndia
 
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E..."Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...ISPMAIndia
 
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdf
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdfJourney of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdf
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdfISPMAIndia
 
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdf
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdfIIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdf
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdfISPMAIndia
 
Careers360_Student Platform.pdf
Careers360_Student Platform.pdfCareers360_Student Platform.pdf
Careers360_Student Platform.pdfISPMAIndia
 
Billgren - Product Summit Speed Layers v12.pdf
Billgren - Product Summit Speed Layers v12.pdfBillgren - Product Summit Speed Layers v12.pdf
Billgren - Product Summit Speed Layers v12.pdfISPMAIndia
 
The Product Manager 2023.pdf
The Product Manager 2023.pdfThe Product Manager 2023.pdf
The Product Manager 2023.pdfISPMAIndia
 
SPM SUMMIT NESL.pdf
SPM SUMMIT NESL.pdfSPM SUMMIT NESL.pdf
SPM SUMMIT NESL.pdfISPMAIndia
 
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...ISPMAIndia
 
Billgren - Value Based Marketing Product Summit.pdf
Billgren - Value Based Marketing Product Summit.pdfBillgren - Value Based Marketing Product Summit.pdf
Billgren - Value Based Marketing Product Summit.pdfISPMAIndia
 
Creating a business case.pdf
Creating a business case.pdfCreating a business case.pdf
Creating a business case.pdfISPMAIndia
 
Storytelling inproductmanagement v0.3
Storytelling inproductmanagement v0.3Storytelling inproductmanagement v0.3
Storytelling inproductmanagement v0.3ISPMAIndia
 
Product roadmap
Product roadmapProduct roadmap
Product roadmapISPMAIndia
 

More from ISPMAIndia (20)

"Scaling Product Leadership with AI" by Ravi Padaki
"Scaling Product Leadership with AI" by Ravi Padaki"Scaling Product Leadership with AI" by Ravi Padaki
"Scaling Product Leadership with AI" by Ravi Padaki
 
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee..."Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...
"Elevate and Innovate: The Art of Product Leadership in the Age of AI" by Dee...
 
Building Products That Think- Bhaskaran Srinivasan & Ashish Gupta
Building Products That Think- Bhaskaran Srinivasan & Ashish GuptaBuilding Products That Think- Bhaskaran Srinivasan & Ashish Gupta
Building Products That Think- Bhaskaran Srinivasan & Ashish Gupta
 
AI Product Management by Abhijit Bendigiri
AI Product Management by Abhijit BendigiriAI Product Management by Abhijit Bendigiri
AI Product Management by Abhijit Bendigiri
 
SPM 2024 – Overview of and benefits of AI in Product Management
SPM 2024 – Overview of and benefits of AI in Product ManagementSPM 2024 – Overview of and benefits of AI in Product Management
SPM 2024 – Overview of and benefits of AI in Product Management
 
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ..."Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...
"Taking an idea to a Product in Health diagnostics" by Dr. Geetha Manjunath, ...
 
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...
Product Manager vs Product Owner – Why Do Companies Still Struggle 23 Years A...
 
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...
AI MODELS USAGE IN FINTECH PRODUCTS: PM APPROACH & BEST PRACTICES by Kasthuri...
 
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E..."Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...
"Customer- Centric AI: Personalising Product Experiences" by Apoorva Gudla, E...
 
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdf
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdfJourney of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdf
Journey of transforming legacy monolith to SaaS_Kartik_Suryanarayan.pdf
 
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdf
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdfIIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdf
IIM BLR Smart Product & Deep Tech Strategy Healthcare 40 BANI world.pdf
 
Careers360_Student Platform.pdf
Careers360_Student Platform.pdfCareers360_Student Platform.pdf
Careers360_Student Platform.pdf
 
Billgren - Product Summit Speed Layers v12.pdf
Billgren - Product Summit Speed Layers v12.pdfBillgren - Product Summit Speed Layers v12.pdf
Billgren - Product Summit Speed Layers v12.pdf
 
The Product Manager 2023.pdf
The Product Manager 2023.pdfThe Product Manager 2023.pdf
The Product Manager 2023.pdf
 
SPM SUMMIT NESL.pdf
SPM SUMMIT NESL.pdfSPM SUMMIT NESL.pdf
SPM SUMMIT NESL.pdf
 
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...
ISPMA White Paper - Ensuring continuous PMF - Hans-Bernd Kittlaus - SPMS Indi...
 
Billgren - Value Based Marketing Product Summit.pdf
Billgren - Value Based Marketing Product Summit.pdfBillgren - Value Based Marketing Product Summit.pdf
Billgren - Value Based Marketing Product Summit.pdf
 
Creating a business case.pdf
Creating a business case.pdfCreating a business case.pdf
Creating a business case.pdf
 
Storytelling inproductmanagement v0.3
Storytelling inproductmanagement v0.3Storytelling inproductmanagement v0.3
Storytelling inproductmanagement v0.3
 
Product roadmap
Product roadmapProduct roadmap
Product roadmap
 

Recently uploaded

Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfAarwolf Industries LLC
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxLoriGlavin3
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentPim van der Noll
 
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotes
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotesMuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotes
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotesManik S Magar
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersNicole Novielli
 
Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Karmanjay Verma
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesBernd Ruecker
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...Wes McKinney
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxLoriGlavin3
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPathCommunity
 
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...Jeffrey Haguewood
 
Long journey of Ruby standard library at RubyConf AU 2024
Long journey of Ruby standard library at RubyConf AU 2024Long journey of Ruby standard library at RubyConf AU 2024
Long journey of Ruby standard library at RubyConf AU 2024Hiroshi SHIBATA
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Alkin Tezuysal
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Strongerpanagenda
 
Scale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterScale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterMydbops
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch TuesdayIvanti
 
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyes
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyesHow to Effectively Monitor SD-WAN and SASE Environments with ThousandEyes
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyesThousandEyes
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxLoriGlavin3
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityIES VE
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality AssuranceInflectra
 

Recently uploaded (20)

Landscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdfLandscape Catalogue 2024 Australia-1.pdf
Landscape Catalogue 2024 Australia-1.pdf
 
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptxUse of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
Use of FIDO in the Payments and Identity Landscape: FIDO Paris Seminar.pptx
 
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native developmentEmixa Mendix Meetup 11 April 2024 about Mendix Native development
Emixa Mendix Meetup 11 April 2024 about Mendix Native development
 
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotes
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotesMuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotes
MuleSoft Online Meetup Group - B2B Crash Course: Release SparkNotes
 
A Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software DevelopersA Journey Into the Emotions of Software Developers
A Journey Into the Emotions of Software Developers
 
Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#Microservices, Docker deploy and Microservices source code in C#
Microservices, Docker deploy and Microservices source code in C#
 
QCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architecturesQCon London: Mastering long-running processes in modern architectures
QCon London: Mastering long-running processes in modern architectures
 
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
The Future Roadmap for the Composable Data Stack - Wes McKinney - Data Counci...
 
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptxPasskey Providers and Enabling Portability: FIDO Paris Seminar.pptx
Passkey Providers and Enabling Portability: FIDO Paris Seminar.pptx
 
UiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to HeroUiPath Community: Communication Mining from Zero to Hero
UiPath Community: Communication Mining from Zero to Hero
 
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...
Email Marketing Automation for Bonterra Impact Management (fka Social Solutio...
 
Long journey of Ruby standard library at RubyConf AU 2024
Long journey of Ruby standard library at RubyConf AU 2024Long journey of Ruby standard library at RubyConf AU 2024
Long journey of Ruby standard library at RubyConf AU 2024
 
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
Unleashing Real-time Insights with ClickHouse_ Navigating the Landscape in 20...
 
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better StrongerModern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
Modern Roaming for Notes and Nomad – Cheaper Faster Better Stronger
 
Scale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL RouterScale your database traffic with Read & Write split using MySQL Router
Scale your database traffic with Read & Write split using MySQL Router
 
2024 April Patch Tuesday
2024 April Patch Tuesday2024 April Patch Tuesday
2024 April Patch Tuesday
 
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyes
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyesHow to Effectively Monitor SD-WAN and SASE Environments with ThousandEyes
How to Effectively Monitor SD-WAN and SASE Environments with ThousandEyes
 
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptxThe Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
The Fit for Passkeys for Employee and Consumer Sign-ins: FIDO Paris Seminar.pptx
 
Decarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a realityDecarbonising Buildings: Making a net-zero built environment a reality
Decarbonising Buildings: Making a net-zero built environment a reality
 
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance[Webinar] SpiraTest - Setting New Standards in Quality Assurance
[Webinar] SpiraTest - Setting New Standards in Quality Assurance
 

"The Transformative Power of AI and Open Challenges" by Dr. Manish Gupta, Google

  • 1. Manish Gupta Google Research India The Transformative Power of AI and Open Challenges Acknowledgment: Work done by several colleagues at Google Research
  • 2. ● AI Trends ○ Deep Learning (dominant paradigm within ML) ○ Foundation Models (new paradigm within DL) ● AI for Synthesis and Creativity ● Inclusive AI ○ Languages ○ Health ○ Agriculture ● Some Challenges and Takeaways Outline
  • 3. “cat” Deep Learning Modern Reincarnation of Artificial Neural Networks Collection of simple trainable mathematical units, organized in layers, that work together to solve complicated tasks Key Benefit Learns features from raw, heterogeneous, noisy data No explicit feature engineering required What’s New new network architectures, new training math, scale
  • 4. Deep Learning Phase 1: A decade of amazing progress in what computers can do Input Output Pixels: Audio: Pixels: “Hello, how are you?” “leopard” “How cold is it outside?” “Bonjour, comment allez-vous?” “A cheetah lying on top of a car”
  • 5. ● AI Trends ○ Deep Learning (dominant paradigm within ML) ○ Foundation Models (new paradigm within DL) ● AI for Synthesis and Creativity ● Inclusive AI ○ Languages ○ Health ○ Agriculture ● Some Challenges and Takeaways Outline
  • 6. Large, pretrained models with self-supervision Example: BERT Obama was born in 1961 in Honolulu , Hawaii , two years after the territory was admitted to the Union as the 50th state . Raised largely in Hawaii , he also spent one year of his childhood in Washington state and four years in Indonesia. v1 v2 v3 v4 .. v11 v12 ….v20 … v24 ... v33 …..v47 Masks (3, 11, 12, 20, 24, 29, 33) Encode Predict masked words BERT Original words Masked words Obama was ____ in 1961 in Honolulu , Hawaii , ____ ____ after the territory was admitted to the ___ as the 50th ____ . Raised largely in ____ , he also spent one year of his ____ in Washington state and four years in Indonesia. born two years Union state Hawaii childhood
  • 8. Chain of Thought Chain of Thought Prompting Elicits Reasoning in Large Language Models, Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou, https://arxiv.org/abs/2201.11903
  • 9. Chain of Thought Chain of Thought Prompting Elicits Reasoning in Large Language Models, Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Ed Chi, Quoc Le, and Denny Zhou, https://arxiv.org/abs/2201.11903
  • 10. Labeled medical images How can we utilise unlabelled images? Large natural datasets Self and semi supervised learning (SSL) Unlabeled medical images Big Transfer (BiT) Chest X-Ray interpretation Breast cancer detection Skin condition diagnosis
  • 11. Gemini Multimodal Reasoning Gemini: A Family of Highly Capable Multimodal Models, by the Gemini Team, arxiv.org/abs/2312.11805
  • 13. ● AI Trends ○ Deep Learning (dominant paradigm within ML) ○ Foundation Models (new paradigm within DL) ● AI for Synthesis and Creativity ● Inclusive AI ○ Languages ○ Health ○ Agriculture ● Some Challenges and Takeaways Outline
  • 14. Transform and simplify the creative process with AI AI image synthesis is the biggest innovation in image making since the invention of photography 200 years ago. - Fredo Durand, MIT
  • 15. “Abstract surreal painting of a melting white 1974 Porsche 911 oozing onto a chess board. Chess pieces are scattered around. A melting golden clock shines brightly in the background. Soft and eery lighting.”
  • 16. simple destructive process slowly maps data to noise [Image credit: Ben Poole] [Jascha Sohl-Dickstein et al., 2015] [Yang Song & Stefano Ermon, 2019] [Jonathan Ho et al., 2020] Diffusion model is trained to map noise back to data Diffusion models [Nanxin Chen et al., 2020] [Durk Kingma & Tim Salimans et al., 2021] [Prafulla Dhariwal, Alex Nichol, Aditya Ramesh et al., 2020+] Google Confidential
  • 17. Model scaling brings consistent quality wins A portrait photo of a kangaroo wearing an orange hoodie and blue sunglasses standing on the grass in front of the Sydney Opera House holding a sign on the chest that says Welcome Friends!
  • 19.
  • 20. Image Extrapolation – Dramatic Uncrop MaskGIT: [Chang et al., CVPR'22]
  • 21. Agriculture Efficiency of LLMs Inclusive AI Information in Indian Languages Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
  • 22. Agriculture Efficiency of LLMs Inclusive AI Information in Indian Languages Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
  • 23. English in India Superset of formal English Variants in query logs User Intent sift dejar price Swift Dzire price boys shuj boys shoes baground piccher background picture dabal bed farnichar double bed furniture mehndi dijain mehndi design IndoPhoneme: Generate phonetic misspellings influenced by user’s native language
  • 24. IndoPhoneme Generate phonetic misspellings influenced by users native language. Tamil Speaker income -> ingam Hindi Speaker shoes -> shuj Corrupted word Correct word Phoneme sequence of the correct word Corrupted phoneme sequence How do we mine phoneme corruptions? Synthetic Noisifiers Round Trip Transliteration Noisy Channel Model Launched as part of Google Search spell correction
  • 26. MuRIL: Multilingual Representations for Indian Languages ● BERT for Indian languages. Two versions: Base (236M), Large (550M) ● Largest coverage of Indian languages (17, including English) ● Supports native and Latin transliterated text +27.8% Available at TensorFlow Hub HuggingFace arxiv.org/abs/2103.10730 +10.1%
  • 27. காைல 8 மணிக ் கு அலாரம ் CREATE_ALARM(DATE_TIME(8 AM)) Multilingual Semantic Parser English alarm at 8 am Urdu ‫اﻻرم‬ ‫ﮐﺎ‬ ‫ﺑﺟﮯ‬ 8 ‫ﺻﺑﺢ‬ Bengali সকাল ৮টায় এলামর্ম Hindi स ु बह 8 बज े अलाम र्म Tamil MuRIL in Google Assistant launched for 8 Indian Languages
  • 28. Proprietary + Confidential M rni: Multimodal Representation for India Morni Vision Build the best multimodal representation for 100+ Indic languages and accelerate development of inclusive and equitable Indic language technologies Population: 1.4B Languages: 60 (1M+), 125 (100K+) Google ASR: 11 languages go/morni
  • 29. Proprietary + Confidential Vaani: Capturing the Speech Landscape in India A given language varies across region, collecting data anchored on language doesn’t capture on the ground diversity Vaani approach Collect image-prompted speech data anchored on region, rather than language, while ensuring diversity 15.4K transcribed 773 districts 154K speech Open Source vaani.iisc.ac.in PoC: partha@, dineshtewari@
  • 30. Proprietary + Confidential LLM Composability go/lm-composition PoC: brachit@ Making LLM reasoning abilities available for low-resource languages through composition Encapsulating org-wide focus areas of model unification, re-use, and modularity. Emphasising minimal model interventions, data and compute efficiency, and flexibility.
  • 31. Proprietary + Confidential Proprietary + Confidential Agriculture Efficiency of LLMs Inclusive AI Information in Indian Languages Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
  • 32. Agricultural Landscape Understanding & Field Event Detection (from Satellite Imagery) Foundational layer of an agri-stack - use cases like farmer loans, crop insurance, incentives to avoid stubble burning, carbon modeling to incentivize regenerative agricultural practices (demo)
  • 33. Agriculture Efficiency of LLMs Inclusive AI Information in Indian Languages Fundamental Research on AI (Robustness, Fairness, Safety, Explainability)
  • 34. Rigidity in ML Models By 2026, more than 80% of enterprises will have used generative artificial intelligence (GenAI) application programming interfaces (APIs) or models, and/or deployed GenAI-enabled applications in production environments, up from less than 5% in 2023, according to Gartner, Inc. …. Analogous to pre-virtualization era when workloads would run on a fixed set of compute resources (e.g., fixed number of processors)
  • 35. Vision: Enable Elasticity of Foundation Models Accuracy Cost Latency Gemini right sized model Without any additional training! Deliver system level efficiencies and business enablement similar to what Cloud did for enterprise workloads
  • 36. Proprietary + Confidential MatFormer: Matryoshka Transformer ● MatFormer builds upon MRL ● Apply MRL to MLP layer in each transformer block L-th layer token x W1 W2 L+1-th layer token x That is, neurons are nested in each other. Model-S Model-XL Model-M Model-L
  • 37. Transformer Block 1 Transformer Block 2 Transformer Block 3 Transformer Block 4 Transformer Block 5 Transformer Block 6 Model-M Model-S Mix'n'Match & Routing on MatFormer ● Mix’n’Match: 100s (combinatorial) of static (on-demand) models for all accuracy-compute ● Routing: Token based routing akin to MoE to realize dynamic computation
  • 38. Matformer: Evaluation 1-shot GPT-evals ● Almost matching accuracy for MatFormer–[XL, L, M, S] models against Baselines ● We get all the intermediate models denotes by ★ for “free” ○ No extra training!
  • 39. Some Outstanding Challenges ● Robustness ● Safety ● Bias and Fairness ● Explainability
  • 40. Objectives for AI Applications 1. Be socially beneficial. 2. Avoid creating or reinforcing unfair bias. 3. Be built and tested for safety. 4. Be accountable to people. 5. Incorporate privacy design principles. 6. Uphold high standards of scientific excellence. 7. Be made available for uses that accord with these principles. AI Applications We Will Not Pursue 1. Technologies that cause or are likely to cause overall harm. 2. Weapons or other technologies whose principal purpose or implementation is to cause or directly facilitate injury to people. 3. Technologies that gather or use information for surveillance violating internationally accepted norms. 4. Technologies whose purpose contravenes widely accepted principles of international law and human rights. Google’s AI Principles https://ai.google/principles/
  • 41. ● AI is here - already impacting every industry and the world at large ● Rise of Foundation Models - models pre-trained on large data, usually with self-supervision ● Huge opportunity to develop AI models and solutions to bring benefits to billions of new people - outstanding challenges Takeaways These are exciting times!