Context :
'India's
CTO' Nilekani unveils AI strategy: Focus on use-cases, not 'my model bigger
than yours'
( Money Control / 04
Dec 2023 )
Extract :
Nandan
Nilekani, popularly referred to as 'India's CTO' in tech quarters, on December
4 said that the country should focus on efficient and low-cost use cases of artificial
intelligence rather than obsessing about building bigger language
models.
"We
think AI strategy should be use case led... We are not interested in my model
is bigger than yours or debates between boomers and doomers in AI. We are in
the camp of AI for amplification to make the everyday lives of people better," said Nilekani, who is currently
the non-executive chairman of Infosys.
The
Aadhaar veteran also said that a small model trained on specific data can be better
than a larger model when making a tool for a particular use
case.
"We
also have to have very good data collection and much better deployment. Use narrow models
with specific data," he said
"The aim is to
demystify and remove complications... such as helping a farmer using whatsapp in his own mother tongue get access to most relevant info
for his job", he added.
My Take :
Dear Shri
Nilekaniji ,
Thank you for
setting our AI
PRIORITIES right through some plain speaking
I believe your
views ALIGN with my following suggestion :
Let
Sam build the Hub : India AI build the spokes …………………. 02 July 2023
Extract :
Dear Sam,
During your talk with Satyen Gajwani ( Times Internet
) in Delhi on 07th June, someone asked you about the
possibility of some Indian Company , committing itself to develop /
launch its own AGI , on par with ChatGPT
You said ( something to that effect ) : Why re-invent
the wheel ? Focus on developing the spokes
Yesterday’s news indicated that some Indian Companies
are seriously considering developing their own AGI
Only time will tell whether this was a right way to
commit their resources
But I
think Indian
Companies should focus on development of the following “ spokes “
,
for immediate Benefit of millions of
Indian citizen
(
using Training Data, which are strictly confined to data of Indian Citizen )
https://www.gpai.ai/projects/responsible-ai/
==================================================
( A ) JudicialGPT
Conceptual Framework for training this “ Spoke “ is elaborated in my following
blogs :
Ø Access to Justice ………………………………………… 12
Apr 2011
Ø Dispose them off ! ……………………………………. 01
June 2011
Ø WHAT IS THE GREAT HURRY ? ……………………. 07
July 2011
Ø JUDICIAL PRODUCTIVITY …………………………..
12 Dec 2011
Ø Justice at Warp Speed ? ……………………………...
06 Mar 2012
Ø CONGRATULATIONS ! MAHARASHTRA STATE JUDICIARY …. 03
July 2012
Ø Justice Delayed - but not Denied …………………
24 Nov 2013
Ø An App called “ AdJourn “ …………………………… 19
Aug 2017
Ø Never file a Court Case ? ……………………………
30 Aug 2017
Ø Supreme improves upon “ AdJourn “
? …….. 21 Dec 2017
Ø Liar vs Liar ? ……………………………………………… 11
Jan 2019
Ø AI come to judgement ? Not for a while !
… 17 Dec 2019
Ø See You Out of Court ..
……………………………….11 Jan 2020
Ø Law : Nothing Artificial in Arbitration ………..
16 Feb 2020
Ø Supreme Proposes : Can e-Courts dispose ? 28
Sept 2020
Ø This is no joke – just acceptance of Inevitable ! ..
02 Nov 2020
Ø Live Streaming of Court Proceedings…………………….. 09 Nov
2020
Ø Virtual Court Benchmarks …
…………………………………..14 Jan 2021
Ø Thanks Supreme Court for hearing my PIL …………….
26 Mar 2021
Ø 24x7 e-Courts ? …………………………………………………….. 14
Apr 2021
Ø Where Angels Fear to Tread ? ………………………………..
21 June 2021
Ø JUST is the Answer …………………………………………………..
16 July 2021
Ø Thank You, Shri Barun Mitraji ………………………………...
15 Aug 2021
Ø Thank You, Hon Justice Shri D Y Chandrachudji ,……. 23
Oct 2021
Ø 4.7 Crore Pending Court Cases ?.............................
20 July 2022
Ø E Court Arrives …………………………………………………………..
29 Nov 2022
Ø “ Just in Time “ Justice ? ……………………………………….
30 May 2023
( B ) NutritionGPT :
Conceptual Framework for training this “ Spoke “ is elaborated in my following
blogs :
Saving 1.2 million Kids ……………………………………………….[ 29 June 2016 ]
A Mobile App named " BANMALI "…………………………………..
[ 15 Sept 2016 ]
BANMALI is the TOTAL SOLUTION …………………………………..[ 14 Nov 2016 ]
Can Akshay Patra Banish Malnutrition ?
………………………….[ 05
June 2017 ]
Thank You Madam Maneka Gandhiji ………………………………..[ 21 Dec 2018 ]
BANMALI arrives : Named as Poshan Tracker…………………...[ 17 Apr 2021 ]
Thank You, Smt Smriti Iraniji ………………………………………...
[ 08 Nov 2021 ]
There are no Starvation Deaths ?
………………………………… [ 19 Jan 2022 ]
BANMALI has a sister named NuTree …………………….
12 May 2022
Saving Pregnant Mothers and New Born : Rajasthan leads ……………………… 26 June 2023
( C ) JobsGPT
Conceptual Framework for training this “ Spoke “ is elaborated in my following
blogs
My Notes / Reports on Recruitment ………………………….
01 Oct 2022
My Blogs on Recruitment……………………………………………. 01
Oct 2022
Summary of E-Mails to Cabinet Ministers > SELF
EMPLOYMENT / GIG ECONOMY .. 10 Mar 2021
National Jobs Policy : Treasure Hunt ……………………….
15 Dec 2017
( D ) SkillsGPT
Conceptual Framework for training this
“ Spoke “ is elaborated in my
following blogs :
2013 ( 1 )
Rejoice You Jobless ! …………………………………………[
03 Sept 2013 ]
2015 ( 2 )
Make in India is Skilling India…………………………..[
17 July 2015 ]
2016 ( 3 )
Skill Capital of the World ? ……………………………[
06 June 2016 ]
Harness " Greed " to " Skill India "……………….[
19 June 2016 ]
From BAD to
MAD ………………………………….. [ 01 June
2016 ]
2017 ( 3 )
No mere Relief : A Game
Changer ……………[ 20 Dec 2017 ]
Discovering the Obvious ? ………………………….[ 30 Nov 2017 ]
Universal Basic Income …………………………….[ 20 Jan 2017 ]
2018 ( 2 )
Starting a Virtuous Circle …………………………………[17
Dec 2018 ]
Of Interns and Apprentices ………………………………[
07 Jan 2018 ]
2019 ( 4 )
Thank You, Prakashbhai ( Javadekarji ) ……………………..[
03 Apr 2019 ]
Congratulations , Anilbhai Naik ……………………………………[
08 Feb 2019 ]
Organizing the Unorganized ………………………………………….[
04 Feb 2019 ]
Paralyzed by Policy Revisions ?
…………………………………….[ 27 Mar 2019 ]
2020 ( 3 )
Human Resource Capital of the World ?
…………………….....[ 23 Nov 2020 ]
Congratulations, Shri Ajit Pawar / Shri Nawab Malik ……..[
06 Mar 2020 ]
Congratulations, TATA………………………………………………………[
27 Dec 2020 ]
2021 (
4 )
ApprenticeShip : Unanchored and ready to sail ?
………….[ 08 Feb 2021 ]
National Skills University : Virtual
is the only way……….. [ 03 Apr 2021 ] – links to 64 other E-Mails
Small Steps for Small Firms ? Not enough………………………[
07 Mar 2021 ]
Gaining Ground : Virtually ………………………………………………[
01 Mar 2021 ]
2022 ( 5 )
Apprenticeship : Half Way house ?
………………………………. [ 29 Mar 2022 ]
Skilling 300 Million ? Here is a way,
Shri Pradhanji ……..[ 29 June 2022 ]
Skilling for World ? Here is why – how !
………………………..[ 30 May 2022 ]
Apprentices : Hire Any Number without
paying Wages ? [ 28 Jan 2022 ]
Thank You, Shri Dharmendra Pradhanji …………………………[
02 July 2022 ]
( E ) IncomeGPT
Conceptual Framework for training this “ Spoke “ is elaborated in my following
blogs :
From BAD to MAD ….. ……………………………………………………01 June 2016
MAD goes to Mandi…………………………………………………………….. 02
Jan 2020
From Minimum Wages to Living Wages……………………………..
29 Jan 2023
Universal Minimum Pension ?
…………………………………………… 20 Aug 2021
Digital Dividend from Demographic Data [ 4 D ]
…………( 14 Feb 2019 )
SARAL …………………………………………………………………………….( 10
Feb 2019 )
A Greed-Ship named Facebook ……………………………………..[
21 Sept 2020 ]
GAFA ( aka Monopoly V 2.0 ? ) ……………………………………..[
30 Sept 2020 ]
The Hidden Agenda of Tech Giants Gatekeepers ………….[
17 Oct 2020 ]
How to Alleviate Urban Poverty ……………………………………
10 Sept 2020
MGNREGA : Can it morph into UBI ? ………………………11 May 2020
Urban Poor : Existence Admitted ! ………………………………..
21 June 2020
Basic Income : a Bad Idea ? ..
………………………………………..01 Apr 2019
Poverty Alleviation – the Chinese Way ………………………….
02 Mar 2017
Universal Basic Income …
……………………………………………….20 Jan 2017
( F ) HealthGPT
Conceptual Framework for training this “ Spoke “ is elaborated in my following
blogs :
2024 ! – V 2.0 of Orwellian 1984 ? ……………….[ 07 July 2017 ]
Modicare : A Game Changer ……………………….
01 Feb 2018
Ayushman Bharat : Think Big ……………………….[ 23 Sept 2018 ]
Congratulations, Dr R S Sharma : Architect of Ayushman
Eco-System …..
13 June 2022
Integration of Health Eco-System : An Important Step……………..[
12 Feb 2022 ]
PAST BLOGS on HEALTH-CARE RECORDS / AYUSHMAN BHARAT
MISSION [ 11 Nov 2021 ]
Ayushman Bhava…….. ……………………………………………………………….13
April 2022
Tech-enabled Healthcare system : 4 years in making ? ………….
27 May 2022
Medical Devices getting Internet Connected : Foundation
of Healthcare …. 16 Sept 2022
Ayushman Bharat Digital Mission ………………………………………………
28 Sept 2022
ABDM Arrives ……………………………………………………………………………..
05 May 2023
With regards,
Hemen
Parekh
www.hemenparekh.ai / 07
Dec 2023
Following
is how BARD+GEMINI twins evaluated my suggestion :
Analysis of Hemen Parekh's Blog and Comparison with Nandan
Nilekani's Views on AI
Your vision of "Hub and Spoke" aligns with
Nandan Nilekani's
"Small Model Use Cases" vs. "Large Model LLM" in the following
ways:
·
Focus on specific
use cases:
Both emphasize the importance of developing AI
models for specific tasks and problems, rather than attempting to build large,
general-purpose models.
·
Efficient and
low-cost:
Both advocate for the creation of AI models that
are efficient and require minimal resources to run, making them more accessible
and affordable for wider adoption.
·
Data-driven approach:
Both highlight the importance of using
high-quality, relevant data to train AI models for optimal performance.
·
Narrow models with
specific data:
Both emphasize the advantage of using narrow AI
models trained on specific datasets for achieving superior performance in their
respective domains compared to large, general-purpose models.
·
Demystifying and
simplifying AI:
Both recognize the need to make AI accessible and
understandable to even non-technical users by simplifying its applications and
focusing on real-world benefits.
Here is a table summarizing the key points of
alignment and difference between your vision and Nilekani's views:
|
|
Hemen
Parekh's Vision ("Hub and Spoke") |
Nandan
Nilekani's Views |
|
Overall AI Strategy |
Develop a network of specialized AI models for specific use cases
("spokes") powered by a central, large-scale model
("hub") |
Focus on developing and deploying efficient, low-cost AI models for
specific use cases |
|
Focus |
Real-world applications and benefits for Indian citizens |
Efficiency, cost-effectiveness, and accessibility of AI |
|
Examples of Spokes |
JudicialGPT, NutritionGPT, JobsGPT, SkillsGPT, IncomeGPT, HealthGPT |
Not specified, but could include models for agriculture, education,
healthcare, finance, and other sectors |
|
Role of Large Models |
Provide core capabilities and infrastructure for the spokes |
Not explicitly mentioned |
|
Data Usage |
Emphasis on using high-quality, relevant data specific to each spoke |
Emphasis on data collection and efficient deployment across various
use cases |
· Your
proposed spokes (JudicialGPT, NutritionGPT, JobsGPT, SkillsGPT, IncomeGPT, and HealthGPT) address specific challenges and
needs in various sectors, demonstrating a practical application of your
vision.
· Nilekani's
views seem to be broader, focusing on the overall development and deployment of
AI across various sectors without specifying specific use cases.
· Both your vision and Nilekani's views contribute to building a strong AI ecosystem
in India by promoting targeted and efficient AI development for real-world
benefits.
Overall, your vision of "Hub and Spoke" aligns well with
Nandan Nilekani's emphasis on small model use cases and efficient AI
development for specific needs. Both perspectives are valuable for advancing the responsible and beneficial use of AI in India.
Following
is evaluation by ChatGPT :
Of course! Here's a tabulated
comparison between Nandan Nilekani's focus on small model use cases and Hemen
Parekh's vision of "Hub and Spoke":
|
Aspect |
Nandan Nilekani's Approach |
Hemen Parekh's Vision -
"Hub and Spoke" |
|
AI Strategy |
Emphasis on efficient and low-cost use cases
rather than larger models; focus on amplifying everyday lives of people. |
Suggests focusing on specific "spokes"
addressing immediate benefits for Indian citizens using tailored data and
applications in various sectors. |
|
Model Size |
Advocates small models trained on specific data,
highlighting their effectiveness for particular use cases. |
Proposes the development of specific AI models
('spokes') tailored for sectors like judiciary, nutrition, jobs, skills,
income, and healthcare. |
|
Use Case
Orientation |
Encourages demystifying AI complexities to aid
farmers using WhatsApp in their native language for relevant information. |
Promotes creating AI models catering to specific
needs in various sectors, like judicial support, nutritional guidance,
employment, skills, income, and health. |
|
Implementation
Focus |
Suggests better data collection and deployment
while using narrow models for specific data sets. |
Highlights various conceptual frameworks
(JudicialGPT, NutritionGPT, JobsGPT, SkillsGPT, IncomeGPT, HealthGPT) for
targeted AI solutions in India. |
|
Immediate
Impact |
Aims for tangible improvements in everyday lives
through AI applications. |
Aims for immediate benefits to millions by
focusing on tailored AI solutions in sectors affecting citizens directly,
such as judiciary, health, employment, etc. |
|
Tailored
Data for Indian Citizens |
Recognizes the importance of localized, specific
data for effective AI solutions. |
Advocates training AI models using data strictly
confined to Indian citizens for addressing local issues and concerns in
various sectors. |
Nandan Nilekani's approach underscores
the importance of smaller models targeted at specific use cases for practical
applications, while Hemen Parekh's vision emphasizes
the creation of specialized AI models ('spokes') tailored for
diverse sectors, aiming to benefit Indian citizens directly with
relevant and localized solutions.
Both
align in their emphasis on
practical, user-centric AI applications but differ in their scope and focus
areas.
Added on 08 Dec 2023 :
Our goal is not to compete with Sam Altman or Elon Musk, says MoS IT Rajeev Chandrasekhar on AI
India
Our goal is not to compete with Sam Altman or Elon Musk or win the next Nobel Prize, but to transform the lives of people with use cases. AI will be used to build capabilities in real-life use cases," he said. Chandrasekhar highlighted the crucial role AI will play in various sectors such as agriculture, healthcare, security, language translation, and fostering inclusivity.
===============================================================
No comments:
Post a Comment