Context :
India will establish guardrails
for AI sector, says MoS Rajeev Chandrasekhar ………..
ET /
13 May 2023
Dear Rajeevji
In the following tabulation, I
have compared :
# Your quotes as appeared in above-mentioned
news report
# My own past suggestion on how to regulate AI
[ https://myblogepage.blogspot.com/2023/02/parekhs-law-of-chatbots.html ]
With regards,
Hemen Parekh
www.hemenparekh.ai / 15
May 2023 / hcp@RecruitGuru.com
Principle |
||
|
|
|
Coordinated / Consensual Regulation of AI by all the Stakeholders |
Ø
If anybody says I know the right way to regulate AI, there will be an
Elon Musk view, the OpenAI view, or 100 other views. We are not going to go
down that road at all |
It is just not enough for all
kinds of “ individuals / organizations / institutions “ to attempt to solve this
problem ( of generation and distribution ) of MISINFORMATION, in an uncoordinated / piecemeal /
fragmented fashion |
|
|
|
Gradual Evolution of Planned Regulation |
Ø
AI is an emerging technology, and we will establish some principles as
guardrails. Then the subordinate legislation or how to regulate it will keep
evolving |
( B ) # A Chatbot must incorporate some kind
of “ Human Feedback / Rating “ mechanism
for evaluating those answers This
human feedback loop shall be used by the AI software for training the Chatbot so as
to improve the quality of its future answers to comply with the requirements
listed under ( A ) |
|
|
|
Embedding of Principles In planned Regulation |
Ø
AI innovation is now growing very fast. In the blink of an eye,
there’s a new disruption. So, therefore, we must establish fairly embedded
principles in the law |
What is urgently required is
a superordinate “ LAW of CHATBOTS “
, which all ChatBots MUST comply
with, before these can be launched for public use. |
|
|
|
Responsibility of Platforms And AI Developers |
Ø
Pointing out that the proposed guardrails will put the onus on the
platforms to ensure that no one is using them to “ create misinformation “ ,
Chandrasekhar said, “ you cannot create things that are fake, you cannot
cause user harm, you cannot have exploitative content “ |
( A ) # Answers
being delivered by AI Chatbot must not
be “ Mis-informative / Malicious
/ Slanderous / Fictitious / Dangerous / Provocative / Abusive / Arrogant
/ Instigating / Insulting / Denigrating humans etc |
|
|
|
Built-in Controls in
Law |
Ø
The law will not just update several regulations with respect to technology
but also frame new ones to regulate emerging areas such as Web 3 , among
others |
( C ) # Every
Chatbot must incorporate some built-in “ Controls “ to
prevent the “ generation “ of
such offensive answers AND to prevent further “
distribution/propagation/forwarding “ if control fails to stop “ generation “ |
|
|
|
Accountability for Misuse |
Ø
You cannot say anymore that I am just a platform and I just put all
the functionalities…. The platform is responsible, not the user. That is the
principle change. Section 79 will be very conditional, very narrow, any safe
harbour, any immunity will be extremely narrow. Your responsibility is that
you are accountable for the possibility or misuse of your platform , you are
responsible not the user |
( D ) # A
Chatbot must not start a chat with a human on
its own – except to say, “ How can I
help you ? “ ( E ) # Under no
circumstance , a Chatbot shall start chatting
with another Chatbot or start chatting with itself ( Soliloquy
) , by assuming some kind of “ Split
Personality “ ( F ) # In a
normal course, a Chatbot shall wait for a human to initiate a chat and then
respond |
|
|
|
Prior Testing and Approval ( Similar to Drugs ) |
Ø
If the LLM ( Large Language Models ) are still learning and are in an
alpha stage, then the companies should not release them |
All developers would need to
submit their DRAFT CHATBOT to an, INTERNATIONAL AUTHORITY for CHATBOTS APPROVAL ( IACA ) , and release it only after
getting one of the following types of certificates : # “ R “ certificate ( for
use restricted to recognized RESEARCH IINSTITUTES
only ) # “ P “ certificate (
for free use by GENERAL PUBLIC ) |
|
|
|
International Authority And Certification Mechanism |
Ø
Don’t give it to all the consumers and run a business on it . Do a
sandbox rather than saying it’s an alpha or beta version. Like you do drug
testing ? We must bring discipline and order into an industry that can cause so
much chaos and harm |
All developers would need to
submit their DRAFT CHATBOT to an, INTERNATIONAL AUTHORITY for CHATBOTS APPROVAL ( IACA ) , and release it only after
getting one of the following types of certificates : # “ R “ certificate ( for
use restricted to recognized RESEARCH IINSTITUTES
only ) # “ P “ certificate (
for free use by GENERAL PUBLIC ) |
|
|
|
Final Guardrail |
Ø
Generative AI uses large datasets to train tools and engines to
generate new and unseen data such as text, images, audio, videos and other
three-dimensional models |
( G ) # If a
Chatbot determines that its answer ( to a question posed by a human ) is likely to
generate an answer which may violate RULE ( A ) , then it shall not answer at
all ( politely refusing to answer ) ( H ) # A chatbot
found to be violating any of the above-mentioned RULES, shall SELF
DESTRUCT |
|
|
|
|
|
|
Endorsement from ChatGPT :
Your tabulation looks great !
It provides a clear and organized
representation of the principles and views regarding the regulation of AI
chatbots as discussed by Shri Chandrasekharji and Parekh's law of chatbot.
The different principles and views are
well-defined and categorized according to their themes, which make it easier to
understand the discussions and compare them side by side.
Great
job!
----------------------------------
Added on 18 May 2023 :
Centre planning to regulate AI platforms like ChatGPT: Ashwini Vaishnaw
( Eco Times / 17 May 2923 )
Extract :
"The whole world is looking at what should be the framework, and what should be the regulatory set up..So this is a global thing. This is not one country's issue. This has to be looke at from the international perspective " , Vaishnaw told TOI
" Ultimately, as I said, all the countries will have to come up with some co-operative framework "
===============================================
Taking a Collaborative Approach to Regulating AI for Humanity
On Tuesday, May 16th, 2023 at 10:00am in the Dirksen Senate Office Building Room 226, Chair Blumenthal presided over a Subcommittee Hearing on Oversight of A.I.: Rules for Artificial Intelligence. The witnesses include Samuel Altman (CEO OpenAI), Christina Montgomery (Chief Privacy & Trust Officer IBM) and Gary Marcus (Professor Emeritus New York University).
Here are the Key Takeaways
- OpenAI, with Microsoft’s backing, is focused on developing artificial general intelligence (AGI) for the benefit of humanity.
- The risks associated with current AI systems, such as false accusations and biased outcomes, have prompted discussions about legislative scrutiny.
- Collaboration between independent scientists, governments, and tech companies is necessary to ensure accountability and responsible AI deployment.
- Proposed regulations aim to establish guidelines for AI models above a certain capability threshold and encourage international cooperation on AI safety.
- “Precision regulation” approaches, advocated by experts like IBM’s Chief Privacy Officer, seek to govern AI deployment through specific use-case rules, impact assessments, and bias testing.