Parishrut Jassal (Founder & CEO @ GovernAI) has trained more than 3500 Indian government officials on AI from Himachal Pradesh to Tamil Nadu.
I asked him:
Do Indian Government officials actually understand AI, or is responsible AI really those nice-sounding phrases that everyone likes to use but without deeply understanding the responsibility behind it?
His answer?
Watch on YouTube; listen on Apple Podcasts or Spotify.
About Parishrut Jassal:
He is the Founder and CEO of GovernAI (https://www.governai.info/), a TEDx Speaker and an empaneled Working Group Member on AI and Digital Governance for the Government of Himachal Pradesh. He has trained more than 3500 Indian state government officials on AI from Himachal Pradesh to Tamil Nadu. Previously, he served as a Deputy Advisor at the Unit for International Cooperation, NIEPA, Ministry of Education, Government of India, where he liaised with Asia-Pacific nations. His Doctoral research investigates the Global Governance of AI for Sustainable Peace, focusing on SDG 16.
Episode Summary:
In this episode, Parishrut shares insights on India's evolving AI landscape, government initiatives, and how states are adopting responsible AI practices. Discover how India is preparing its public sector for AI challenges and opportunities. In this episode, Parishrut shares insights on India's approach to AI governance, model procurement, open source versus closed source models, AI sandboxes, and the unique Indian strategies in AI policy. Discover how India is shaping its AI future with a cautious, inclusive, and sector-specific approach.
Timestamps:
00:00 AIR Bites (Precap)
01:46 Decoding India's AI Governance
02:15 Do Indian Bureaucrats REALLY Understand AI?
08:58 The Data Trick That Shocked Government Officials
10:32 Agentic AI Has Reached Government Offices
13:16 Does India Have a Chief AI Officer Yet?
15:11 Claude Mythos & Project Glasswing: Is India Ready?
16:22 The Indian States Winning the AI Race
19:35 South vs North: India's AI Divide
22:33 Naming India's #1 AI Role-Model State
24:44 How States Are Secretly Coordinating on AI
26:41 India's Central Hub for AI Knowledge-Sharing
29:20 GPT, Claude, or Sarvam? How Govt Buys AI
31:51 Open Source vs Closed AI: What Govt Prefers
32:50 Inside India's AI Testing Sandboxes
33:22 Does India Have an AI Use-Case Database?
37:21 Is There an AI Incident Reporting System?
40:27 The "Indian Way" of AI Governance
44:06 India's Biggest AI Governance Blind Spots
49:03 How YOU Can Break Into AI Policy?
52:00 Free AI Governance Courses & Fellowships
57:25 Closing Thoughts & Thank You
01:00:10 AIR BTS
01:00:32 End Credits
Transcript:
Aashka Patel (00:04)
hello and welcome to On Air with Aashka Welcome to the show, Parishrut. Finally, we are here. Thank you so much for joining us today. Yeah. So, let’s dive
Parishrut (00:12)
Thank you, Matt.
Aashka Patel (00:13)
right into the questions. you have trained the Indian state government officials on AI from Himachal Pradesh to Tamil Nadu.
Do they actually understand AI or is responsible AI really those nice sounding phrases that everyone likes to use but without deeply understanding the responsibility behind it?
Parishrut (00:33)
Thank you so much for your question. I think... From my POV when I started...
started this intervention and going state to state and started engagement with the stakeholders, right? Stakeholders for I can get my share of contribution to the public sector from the responsible AI point of view. So I believe that the understanding and the kind of engagement we used to have in the
Aashka Patel (00:57)
Yeah.
Parishrut (00:57)
initial phase has tremendously changed now. So like earlier it used to be artificial intelligence. So it would always be
Chat GPT or any other
synonym any other familiar word they can understand few of them
were able to caution me like okay that cleaning robot
Aashka Patel (01:19)
Hahaha
Parishrut (01:20)
also has AI so is that also Chat GPT and some kind of like that
that’s very innocent but genuine question genuine curiosity i would say and then now
onwards now how i see that like even our sessions have also increased
Earlier we used to cover from the perspective of how artificial intelligence has evolved to what we see as of now. Starting
from the back year of 1956 and how the term coined and how certain industry sectors were affected by it earlier. But now
if I see regular words like responsible AI, regular words, regular use of words like hallucination, biasness,
Mostly they can do understand what is the connotation of the word from
like, know, what it in generic, like day to day life, what it means, but
they are able to bring it into the picture, bring it into the
interplay when we are having a conversation about, okay, if AI can be solution for any public sector problem at the end of the day. So from that
point of view, would say, and if I have to really see from my experience with different state governments,
It’s
not much but as much as we have been able to garner, I would say that it depends upon the department
irrespective of the state.
Aashka Patel (02:45)
Huh.
Parishrut (02:46)
In as you would say, in all the conferences we used to catch up earlier, like two years back or one year back, we would see that even in the central government landscape, the MeiTY
had a better understanding with respect to AI governance, other than like any other, like, know.
Aashka Patel (03:09)
Yeah.
Parishrut (03:10)
address of the problem used to be reactive, I would say. Because
like if any AI model or LLM model is like creating a chaos or
menace to the society, then not
only such solutions were taken into consideration. But now I would say that even at the central government, we can see different line of ministries have like really revamped their staff,
revamped their policies with respect to artificial intelligence.
In the same way in the state governments I would say in the
In a department, if the top bureaucrats, level three bureaucrats as per the competency framework layout by MeiTY itself I would say they
have a better understanding with respect to how
sensitive artificial intelligence is and how carefully we have to be with it. And they have catched up with,
the jargons I would say, but also the
important keywords because
that only you can like, know, really tap into the problems and understand and bring into the other stakeholders from the technical side or from the
legal side or from the policy side, right? So
in the same department, I would say the
fresh officers who are young, who are just joining or who are mid-serving officers, they
are still, having a hard time to understand because...
Aashka Patel (04:31)
I understand,
Parishrut (04:33)
Yeah, because there are other crucial
and time sensitive issues they have been addressing in different departments.
Aashka Patel (04:40)
And yeah,
of course, at the end of the day, they are policymakers or like non-technical people trying to catch
Parishrut (04:45)
Yes.
Aashka Patel (04:47)
up with the tech. So, of course, it makes sense.
Parishrut (04:50)
Yeah,
so I would say on a concluding note that.
The IT department at each and every state have been to revamped and in well consideration of the proof of artificial intelligence adoption and
what kind of like solutions it can bring to the public sector, public service delivery.
And from the capacity angle, I would say they do understand that’s why we keep on engaging with them because they see the need of like, know, keep
officers trained or like, revamped about it.
other than that, now I see a trend of, you can say, a ripple effect because all the departments
are feeling that we should also know about it and from the right angle
That’s why we need more of the interventions, the contributions from the public and private partnerships with respect to that.
Aashka Patel (05:47)
That is good. So, you spoke about the evolution in their understanding. tell us deeply about a specific conversation.
When you speak with the IT department, let’s say for example, about a specific problem, what is the starting point like? Let’s use AI to solve this or can AI actually help us solve this? What’s the starting point like?
Parishrut (06:10)
I mean, as I can share this thing from my lens,
initially the conversation used to be, okay,
gentlemen, you can just tell us about what the technology can do to,
like, fix some obstacles we are facing.
I used to walk them through and they
used to really excited about it but they really
wanted to wrap their head around other components with respect to adoption of artificial intelligence. They
were like, what kind of budgets we have to see?
Would it work or is it any other technology solution which is coming into the market? was like, guys I don’t know if it is a technology solution but I think I can bet
on
if we really unpack it and see where we can use it and up to what extent.
But now,
the conversations have really, really paced. Now, they directly comes to,
the tenants of responsible AI, how this solution will map out and what kind of stakeholders we have to really engage. And I’m impressed. it’s a
bit of excitement also to look forward to such conversations and where I can contribute to it. So I believe that that’s the shift I’ve seen.
Aashka Patel (07:21)
Aha, yes, yes,
That’s nice, So, like what’s the most pleasantly surprising thing that you have heard about how they think about AI?
recently,
Parishrut (07:36)
I think recently
So the most pleasant thing I saw is that when I, in one of our sessions,
And then in one of our meetings with respect to the problems which we were just brainstorming in
a couple of states, they were really surprised to see the kind of data presentation we were able to do.
And that was only possible because of like, I’m not a big on like, know, using Power BI tool or any other tool.
kind of like visuals or kind of like ideas you can get from the
existing open source, materials, experts
in Power BI has used it and how they have,
extrapolated the data and presented it, right? In the same manner, we have used like the data.
It’s a simulator basically. Exactly. How you
Aashka Patel (08:38)
Warehouse or anything?
Parishrut (08:44)
can present a scenario. Yeah. And they can see
Aashka Patel (08:45)
visualize.
Parishrut (08:47)
who will be the people who will be impacted, which
segment of the society will be impacted, what are the resources we dealing with. And they can watch it at play.
Maybe that’s randomization or maybe if we can put certain,
guardrails to it and it will outplay or present the outcome
in a span of like a month or so with
certain resources with certain purpose and they were like that’s really interesting and they
were really able to like in the small frame of time come to a better judgment or better
engagement with the solution or with the problem at hand.
And it saves time and
the best part is it’s not even about time. They are ready to give us time to engage with
private sector players and to engage with solutions which the private sector or AI companies in India are bringing to their doorsteps. But
they can really see
the efficiency or how well they are able to grasp the solution or the harms it can bring. So I think
that was very shocking for them. So they were like, how you can do it? Do you know how to code or
something like that?
Aashka Patel (10:05)
So, us go little deeper. like when states or state officials talk about AI, are they still focused on the generative AI
or are they already talking about AI agents and agentic AI?
Parishrut (10:26)
I would say, agentic AI they are excited about it because they can understand.
And as far as the adoption is concerned, agentic AI
systems at a very experimental level, I would say.
Aashka Patel (10:38)
Yes, of course, yeah.
Parishrut (10:40)
And other than that, the basic, to have the readiness
at their,
respective departments with respect to agentic AI also they are excited for that and
they are they keen to adopt it but obviously with a certain precaution and
like sense of understanding
what kind of strength we have of the officers who will be engaged in that, who will be responsible for what role, and
have the knowledge base and competencies to understand if things are not looking okay on screen or maybe
on the...
on the back end, right? So
even in the IT teams, I’ve seen that they are more keen and they have been engaging with
different stakeholders from different organizations and they really would
want to put it on the map for them to make a better decision, sound decision. But as far as the basic adoption, would say at the interpersonal level, it’s easier. Yeah, but
to...
majorly call it as a solution which any of the organizations adopted. There are a couple of,
but we still have to see the right outputs which we have laid out if it is bringing the areas able to get into that.
Aashka Patel (12:00)
Yeah, yeah,
yeah, make sense, make sense. So, you spoke a little bit about roles do all the IT departments and state governments have a chief AI officer role yet or have they appointed it yet or like are they in the process of or where are they?
Parishrut (12:19)
The Nomenclature might not be there but they have the Nodal person assigned already.
Aashka Patel (12:21)
Okay. Okay.
Okay. Okay. Got it. Okay.
Parishrut (12:25)
They have the nodal persons who have
the certain understandings and they are
the people who keep on engaging with us also and they
keep track with what is happening for what the next meeting has to be because
there should be a bridge and
Aashka Patel (12:43)
Yeah, yes.
Parishrut (12:45)
even with interdepartmental stuff, right? Because the IT
department has to be the backbone of all the departments at the end of the day.
Aashka Patel (12:51)
Yes, yes, yes.
Yeah, yeah, yeah.
Parishrut (12:53)
You can’t
like outsource outrightly.
Aashka Patel (12:58)
So, let us come to Claude Mythos. So, after the Claude Mythos discussion and Nirmala Sitharaman’s meeting, how is India preparing its public sector systems to defend against Mythos level attacks? Do you have any insight into that?
Parishrut (13:11)
I
think I would want to put
it in a perspective
Aashka Patel (13:18)
Okay.
Parishrut (13:18)
that we see the capability of the Mythos model. So
it’s not about like, okay, we have to save ourselves from it.
Even the finance ministry’s
request for the early access was from the point of view that we would want to know at what level the threat is
Aashka Patel (13:31)
Hmm. Yes, exactly. Yeah.
Parishrut (13:35)
and at what level the competition level is because we have already seen that
with other LLM models which are open source and which you can
use it at your disposal and like see how what you can
do with it have a Eureka moment or
for whatever right but with this Mythos I would say that it’s it’s more or less
I would say the anthropic won’t agitate us right because
they are saying that no one is prepared as prepared as everyone could be but I would
Aashka Patel (14:07)
So, did we
get early access or not like do.
Parishrut (14:12)
I have to check for it. Like I’m not sure I didn’t follow up on that. But I would
Aashka Patel (14:14)
Because under project glasswing, yeah, it’s okay.
Parishrut (14:19)
say that even if we do get the access, it’s not the main focal point. I would say
it’s the perspective of the Mythos even if it is that harmful. But at the same time,
it can detect vulnerabilities
in any systems. And it has the computational powerful
Aashka Patel (14:36)
Yes.
Parishrut (14:40)
abilities if I have to simplify then we have to see from the tangent or from the aspect
the autonomy these kind of models when if we
instilled in an agentic AI system or for any other purpose or in any
in any framework we
have to really see
how we can contain it at the end of the day. So I think Anthropic
has also gone with their, I’m just
from, I’m forgetting from top of my head, the name. Yeah, yeah, yeah. They have been,
Aashka Patel (15:11)
project glass wing? Yeah,
Parishrut (15:18)
the flag bearers, I would say for AI safety and they have
walk the path as of now and
And the project holds the same promises and we have to see how we can really make that instrumental and understand what Mythos
really is all about.
Aashka Patel (15:37)
Yeah,
so currently what Dario Amadei has been saying is like under project glass wing
Small number of companies have gotten access to Claude Mythos and then they are like the cybersecurity experts, engineers, are using Mythos to detect the vulnerabilities and also to make the systems more secure and robust. So that is why I wanted to know like, did we receive access to that or not? Yeah, no worries.
Parishrut (16:06)
No, not of my knowledge. But I’ll apprise
Aashka Patel (16:14)
So, us move on to the next question. Which Indian state do you think is leading on AI adoption today? Explain any specifics you can.
Parishrut (16:24)
I think there are a couple of states which are
doing surprisingly well, I would say, because if I really track down AI governance journey of India, we have been making decisions at the right time about right things.
And we came a long way. And for the states, mostly,
the southern states have been up to the game. They have, they understand and the kind of adoptions and the kind of resources is required to actually make AI functional to solve problems of the society at the end of the day, right? And
they have been, I would say, prepared for it.
I must say that because as
Aashka Patel (17:17)
You’re okay.
Parishrut (17:19)
you must have like followed up like couple of ministries in the south have like
dedicated now they have a dedicated like you AI ministry earlier
few of the states have AI policy documents readily to guide them.
you can see the, even in the chronology, if we follow up, first we have the policy which has to be in the place because we have to map out all the people are concerned. Then only we have to bring the central nodal agency at the state level who will be looking after who will be thinking from the angle of compliance liabilities and solutions, budget allocations and
the functioning of it, right? So I
think the portfolio, if it is dedicated to a certain level, because the way you ask, if the importance of an AI officer, even at a department level, is a major role. In the same way, I would say that the southern states are trudging towards a promised path.
by
having a dedicated ministries. I
would say that even in the Northern India, the states are following up. Many of the states also do have the AI policy. But we have to see if we can have a dedicated AI wing
at the ministry level. And apart from that, I would say the better thing is which will be one of the parameters to assess.
To answer your question, what
kind of engagements we have from the people, from the public side, from the private sectors, the tech companies of India and from the
regulatory bodies, how
well they are.
how well their input is taken into consideration. think we can see the sectoral regulations are already floating around and
it’s high time now and I look forward with the same excitement with that.
Aashka Patel (19:22)
So
you mentioned the southern states in terms of AI governance, So are they also ahead in terms of AI adoption?
Parishrut (19:32)
I think for the AI adoption, would say that even Maharashtra came up very long with their...
Aashka Patel (19:41)
They were the torch bearers.
Parishrut (19:44)
I would say that they were torchbearers when the time was for their decision, right?
Aashka Patel (19:50)
Okay,
makes sense.
Parishrut (19:54)
the Indian AI governance landscape like each and every stakeholder which is engaged, they have been
taking the decisions in a very timely manner. And
we see a certain
certain disturbance in the social mileau But
that has been catered and contained and
certain guardrails has been placed with respect to that. But if I have to say
that even in the case of Maharashtra I was stating that agriculture sector is something which is
a...
A good example of how from the IT department, the brainstorming or
discussions about artificial intelligence has moved to altogether a different sector. And with respect to
Aashka Patel (20:31)
Sector. huh, yeah.
Parishrut (20:36)
Tamil Nadu and Kerala, I would say they have
the dedicated AI ministries now and that’s a good, I’m worth appreciating, big
step.
Other than that, the other state governments in the southern, they have,
they are already ahead of the game. They have right investments in town.
They have right, players already engaged. They have right conversation in the right rooms. So I think we
have a lot to learn and the rest of the India is really catching up and we have
to see where the north
India altogether from the west to the east has to engage from the same
intensity and interest I would say.
Aashka Patel (21:23)
Hmm.
So if you were to pick one exemplar state like as Sharma Ji ka beta, which one would you pick?
And also
Parishrut (21:33)
I think the connotation of
Aashka Patel (21:34)
mention the reasons like behind it like okay, these are the good stuff. yeah. Okay, yeah, yeah, it’s your judgment.
Parishrut (21:36)
yeah, definitely. I’ll reason with my judgment.
So I would say that the connotation of anyone being the top of their class or top of their work process and workflows, the state which has really been under my radar from quite a long time.
was Tamil Nadu.
Aashka Patel (22:06)
Okay.
Parishrut (22:07)
Yeah, think I have been, specifically because of the IIT Madras and even a
lot of developments I have seen how CeRAI has done a lot
of like, know, contribution to the Indian AI governance landscape, not only limited to any specific
Aashka Patel (22:23)
Yes, yeah.
Parishrut (22:26)
state. So to let educational institution which falls under our HEIs
sprout to a level and having the right access to the resources and engagements and giving time to it. So I think they have really been under my radar so far and but it’s a competitive landscape. I believe yeah any
Aashka Patel (22:52)
Hmm, yeah, an evolving one,
Parishrut (22:55)
Indian families
son or daughter has to be up their game even if they’re getting to IITs or IIMs and I think they have to be on top of the game if they’re a civil servant and they have to top of the game in any walk of life I would say. So the journey is long I would say let’s see who keeps up with the promises.
Aashka Patel (23:11)
Yeah, yes,
yeah, yeah, makes sense. So, let’s get into coordination in terms of state governments. So, are Indian states coordinating in terms of AI adoption or governance and if yes, then how is it actually happening?
Parishrut (23:30)
I would want to walk you through this way
at a very milestone mapping of this because when if I see myself when I ventured into this field and how I have like tried to wrap my head around what is happening and how like you public sector bodies can adopt and like make sound decisions and bring solutions to the public at the end of the day.
you really bank on the other stakeholders also who are engaged, right? So because it’s
not a one person journey or one entity journey, It’s a collaborative effort where we have
to come forward and like, know, really put India at the face of global south at first, with
respect to artificial intelligence regulations and compliance and adoptions,
Aashka Patel (24:20)
Yeah.
Parishrut (24:21)
I would,
Interdepartmental meetings happen, right? And
they have been channelized through the state government, right? But mandate of these meetings are restricted to the state level agendas, right?
Aashka Patel (24:35)
yeah.
Parishrut (24:35)
But from the central government’s POV, I would say that all the Apex officers,
are different nomenclatures.
by the state government, be it our chief secretary or be it any other officer representing the state. They have been in engagement since like, know, 2024 and they have their own internal meetings and
they have the discussions, right discussions to set the right tone for their respective state and collaboratively for the India altogether, right?
And at the union level, as I have discussed that if they are able to facilitate certain conversations from the representative of all the states,
we see a wave of
same adoption of a systematic engagement at the state level also at the end of the day. Right.
And the central government has never shied away from like, know, putting right information, right guidelines, right voice to issues. And that
has been the guiding light for all the states, I would say.
So I think
Aashka Patel (25:41)
That’s how coordination is happening right now.
Parishrut (25:43)
Yes, has been happening.
Aashka Patel (25:46)
Got it, got it. So like for coordination, there’s no place where they can share their learnings like, okay, one state government has deployed something in public sector AI and it hasn’t worked out and they have.
documented their learnings for other states to learn from and then develop on top of that. Like, is there any such system?
Okay. Okay. Okay. That is the central repository for, okay. Got.
Parishrut (26:09)
So, there is a platform, think IndiaAI’s website has been very actively putting on. And now
has been addressed through India AI Mission altogether, has done a tremendously good job.
Aashka Patel (26:40)
Yes, yeah. Hmm.
Parishrut (26:45)
And because if you really see at the end of the day, although we as a company have been catering and addressing to the capacity building or the literacy front as of now. And with our engagement with the public sector, we have seen now the conversation has moved from what AI is to how to procure it.
if India AI Mission has furnished
the right amount of information, which has
been reaching to the right set of people who are interested, their vested interests were here, and who
wanted to join India in this journey of artificial intelligence. Otherwise, the Indian landscape is very big. have a different set of like, you
Aashka Patel (27:31)
Yeah,
Parishrut (27:32)
opportunities and resources and positions to really look forward to but artificial intelligence has
been addressed through India AI mission very aptly AI Kosh is one of the recent example of that because if
you can have the access to through the GPUs which the government has been offering if you can have that
access at a very nominal rate
Aashka Patel (27:57)
year.
Parishrut (27:57)
like that’s brilliant because earlier India had had this infrastructure but it was only limited to a few
of the key eminent technology and other educational institutions in India but now it’s out there it’s out there accessible to the young generation who are building for India and
who are deploying in India.
Aashka Patel (28:20)
Yeah, yeah, makes sense. So, you spoke about procurement. So, let’s talk about that. When government agencies in India implement AI, how do they usually approach model acquisition? Do they primarily procure frontier models like GPT or Claude or are they now looking toward Indian models as well?
Parishrut (28:40)
I think Sarvam is a good example of this. I have
followed them since they have started the engineering and
Aashka Patel (28:48)
Inception.
Parishrut (28:51)
I had my own reservations about to showcase Sarvam
at one of my workshops at the initial level but now I have seen the response from people how well the Indian models have been doing and that has been only possible because
Aashka Patel (29:06)
Yeah.
Parishrut (29:09)
option or the procurement or even making the right space for an entity to flourish happens because as I said it’s a long journey and we have started on this voyage and
let’s see where we get into that.
Aashka Patel (29:24)
Yeah. So is there a preference for open source models or closed source models, or is it a mix like when it comes to public sector AI?
Parishrut (29:33)
With respect to public sector as far as the brainstorming is concerned or at a very experimental level, open
Aashka Patel (29:39)
P. O. C. Level, yeah.
Parishrut (29:41)
source models are very, very like, know, resource efficient, I would say. But
as we are to scale, we have to bring certain guardrails and certain decisions with respect to adoption of artificial intelligence. And that’s
a narrow lane we have to navigate.
Aashka Patel (30:00)
So, are there any AI sandboxes where public sector AI can be tested before deployment and if yes how mature is it?
Parishrut (30:11)
This is very good. This is very good question. would say a sandbox or a framework or a scenario
where you can really see
At what scale we can deploy this artificial intelligence? So you can map the whole lifecycle of any AI solution at the end of the day. And starting
from AI development phase and then having the right information
to pass to the AI assessment stage and then to AI deployment and then to diffusing it to different sectors I would say. In that, RBI’s free guideline is very
very on point because, is one of the sectors which is
a very crucial sector I would say because the number which we will be dealing with and the kind of data we will be handling is not
Aashka Patel (31:06)
No.
Parishrut (31:13)
only holds the characteristics of volume but also of sensitivity.
So
a right framework which has been laid out by RBI do show and do work to any interested parties or any interested entities who would want to build solutions for the banking sector for that matter. In the same way healthcare sector has their own framework which is has an
acronym of SAHI.
And as the word connotates that’s the most rightfulness we would want for a right for a sector that
Aashka Patel (31:45)
Yeah, that’s
Parishrut (31:50)
level not of level but of that importance to the Indian masses because healthcare has been
an old backbone of the whole country because you
see the population
distributed across different jurisdictions of India also
has to be catered with the right set of like healthcare amenities and if AI can contribute to it with the right approach
Aashka Patel (32:16)
Yeah. Yeah.
Parishrut (32:19)
I think let’s go for it right.
Aashka Patel (32:21)
Hmm. Yeah, Healthcare is the major sector to be disrupted with AI. yeah, we do have.
Parishrut (32:27)
Yeah,
these frameworks can be referred as a framework which is a reference point for a solution to be catered to a sector.
So
I think that’s the nascent stage sandboxes we can have. different organizations have been in the process of ER cells have been brainstorming for different sectors, for the banking,
and even for the agriculture, and
for the rural development, and for the urban planning. We have been brainstorming with different frameworks
for our reference point to have an idea of what kind of guardrails we have to keep in mind and what
kind of like interventions we can offer to what kind of like frameworks which we are referring are they
curated in India or
do they align with the global south scenarios. So we
have been taking a few firm steps in the right direction,
not as of now we can share with you as of now. Right. But the approach
of simulators is also in heading in that direction with respect
to.
work at Govern AI and
we do see industries have been using this instrument of sandboxes to
have a fair share of advantage to envision
how AI or any other technology will unfold once deployed.
Aashka Patel (34:10)
Yeah. Makes sense. So does India have proper use case repositories for public sector AI or is like the India AI mission website that you pointed out like is that the one source that people can point to for like private sector and public sector both?
Single
truth available. Yeah, that’s the right word. Thank you for guiding me.
Parishrut (34:36)
So I would say if I have to see the prism of truth we have to address all the artificial intelligence developments.
At the central level, the India AI is the gateway. And then you
have different verticals of it. It has the
study vertical where you can find case studies. It has
to read
Aashka Patel (35:00)
you
Parishrut (35:01)
and to understand from a perspective of a curious reader, I would say. So they have the
blogs presented. And then if I’m not wrong, then we have the right frameworks for the right sector.
which are also laid out there.
Aashka Patel (35:19)
Okay, it’s good.
Parishrut (35:22)
And then it also obviously leads you to the respective domains and respective aspects. But that’s the gateway I would say.
Aashka Patel (35:30)
Yeah, makes sense. That’s something similar to what is happening in the US as well. Like they also have those sector specific repositories like, okay, for healthcare, these are the use cases where we have deployed public sector AI. So it makes sense. So hmm.
Parishrut (35:50)
And even in just to add two bits more
to this, even at the larger AI adoption market, even in other foreign jurisdictions,
would say that the most of the work which is to ensure the
centralized knowledge base incident reports
and what kind of regulation developments are happening has
been catered and done by
like international education institutions. you do see
the academics plays a huge role in this. And
they have been playing the role of a watch dog at the end of the day. So that’s why we need civil societies and these academic institutions also on the table of
discussions with respect to AI at large when it is at discussion.
Aashka Patel (36:43)
hmm
Yeah, makes sense. also like to add to that, like more importantly, these use cases repositories are for AI transparency, to be as transparent to the public as possible with the citizens. And you touched upon incident reporting a bit, but I would like to dive deeper into that. Like, do we have at India clear systems for AI incident reporting, like an AI incident data
base and redressal mechanisms or are they a work in progress?
Parishrut (37:16)
So
AI governance guidelines, which was introduced in full-fledged.
last November, would
say they have like really laid it out that
the mechanisms, the incident reporting mechanism is the right approach to address to it, right? And then we see as at the India Impact Summit, that connotation and the terminology which was used specifically was for the impact, right?
Because if you see what the French has, how they have addressed with respect to artificial intelligence, were giving it the right
voice and if
it is the talk of the town then how we are addressing the impact of it right so I would say that with respect
to that the right platform I would say we are like even we as a civilians can report it I won’t say that that is already in place but
Aashka Patel (38:04)
Yeah.
Parishrut (38:09)
I would want to apprise that now we have discussions happening at the right
departments, which is the
governance and economic advisory group, which is formulated Because
the Office of Principal Scientific Advisor and the institutional ethos of NITI Aayog has really cleared clouds for us to see what artificial intelligence is. Back in 2018 2019 NITI
Aashka Patel (38:37)
Yeah.
Parishrut (38:41)
Aayog has been up
up there talking about this and have been addressing and now
Aashka Patel (38:46)
Yeah.
Parishrut (38:47)
it has been instrumental to for MeiTY to really take it to the further level. Now we need all the right stakeholders which has been formulated and the information is out there
and the right people are being engaged but
obviously at the level of like public service delivery we are seeing and from
the lens of any citizen or from any listener
to this conversation as well. I would want
to say that we would speculate that if certain platforms are already there the same way we see with context to the certain work in cyber security. We would want to see like AI incident reports or
like how AI guardrails are in place or certain toolkits which are
as of now in place but if they are out there. But
I would say
that
from my personal experience I have
served for a very short time with the central ministry as well and what
I have seen is that the kind of information and the kind of solutions what the government has been offering and building
it’s also about the reaching it to the right people in a manner they can understand.
Aashka Patel (40:05)
Yeah,
accessibility is important.
Parishrut (40:09)
Yeah.
And I would say that there are apps for
Aashka Patel (40:13)
you
Parishrut (40:13)
different platforms from different sectors. I think it would be good to see something like that.
Aashka Patel (40:22)
Yeah, at least right steps are taken in that direction. So of course,
Parishrut (40:26)
Yeah.
Aashka Patel (40:27)
we will sooner or later see something happening, concrete happening.
So yeah, what is India doing differently on AI governance that feels distinctly Indian and not just the so-called Washington effect critics keep pointing to apart from creative acronyms like Sahi and other acronyms. So what is one thing that India is doing differently?
Parishrut (40:55)
I think
Washington has been a guiding light at the initial of post-COVID times, right? When artificial intelligence
Aashka Patel (41:03)
Yes. Yeah.
Parishrut (41:07)
came into the picture and even I worked with certain agencies in Washington at
a time where we really prepared reports and assessed countries readiness with respect to artificial intelligence adoption, right?
And...
It’s one thing to put everything on an index, but
we have to see every sovereignty has their own journey to let the technology be used for betterment of the society at large.
Aashka Patel (41:41)
Yes.
Parishrut (41:42)
Because technology doesn’t have any boundaries at the end of the day. That’s as it is, right? And if we have to really cater by assessing the betterment and needs of the society, which we will be serving because it’s one thing to bring a powerful tool to,
for the betterment or for the right reasons, but it’s
Aashka Patel (42:04)
him.
Parishrut (42:05)
altogether a different ordeal to
really able to contain and utilize it optimize it and see the results yeah exactly exactly so so
Aashka Patel (42:13)
To bring the impact. AI impact, might say.
Parishrut (42:21)
and as far as the india what we are doing differently so from a personal perspective when i really
started getting into this domain was when when I see European Union taking for me my daily briefings with respect to what is happening in AI because after some time like not much was happening in Washington with respect yeah yeah I mean they they said what they have to say and now
Aashka Patel (42:51)
With respect to governance, European Union has been leading.
Parishrut (42:58)
I would say we have to really see how we have to, utilize it. We have to it. And for that, think European Union has,
Aashka Patel (43:04)
The Indian way.
Parishrut (43:10)
has been the brunt for it also because they haven’t right outly vocal about it. They had
Aashka Patel (43:17)
Yeah.
Parishrut (43:19)
the infrastructure
to bring something like
a governance structure which is formalized, EU AI
Aashka Patel (43:23)
Act.
Parishrut (43:28)
Act, right? And that’s not the first time the Europeans have able to address it, but how well they have addressed with respect to artificial intelligence, that’s a different altogether talk. But I would say with respect to digital infrastructure, they have like really sorted out with respect
to...
the granular level governance structures you would need from the law perspective, from governance perspective, from the policy perspective, right engagements, right resources, resources beyond monetary aspects. So
Aashka Patel (44:00)
Yeah.
Parishrut (44:02)
those are the things I would say has been like impression of the jurisdiction.
up north but if India has to be addressed with respect to these two entities these two jurisdictions how well they have addressed
it I would say that we have picked a very caution plus informed choice when it comes to artificial intelligence not jumping into the bandwagon of like
Adopting AI or governing AI? Both the things. So we took our time.
And I think like, don’t have like, I don’t want to label
that it comes into that it can be mapped through a research papers perspective.
Okay, so this is what India now falls into.
Are we going
to be the leader for the global south or not? Are we taking the right step or not? Obviously, we will be the
key players in the global south to set the tone for the artificial intelligence governance.
Aashka Patel (45:12)
Yeah, so we have something called scale that we need to deal with. And that I think in the near future, we would be able to show it like, OK, at this massive scale, can.
adopt AI and govern AI with the right intentions. So that would be something that would be entirely different or something unique about India. yeah, so I know this is still evolving, but what are the most urgent gaps in India’s AI governance that from your insider perspective needs attention right now?
Parishrut (45:52)
I think first and foremost I would say making existing frameworks in India which if we map across sectors one thing which is lagging is making them instrumental and how we have to make them instrumental so see industry any industry at the end of the day
few of the mentors who have been guiding light for me as well right because I’m
still learning and growing
Aashka Patel (46:23)
Okay.
Parishrut (46:24)
what each and every mentor or
a person or an organization which I look up to what would be their stance with respect to this
whether it is private or public, it doesn’t matter. If someone is talking sense like, let’s hear them out, right? One thing which is common is
how the economics will play out.
Because artificial intelligence is something if you have to have under the garb of indigenous, it comes at a price,
right? And a price which
can be
catered by a legal tender or
by right engagements, by right
reinforcement of right issues, right values. And
for that, I think the Office of Comptroller Altar Journal has
really played a significant role because
When economics come to the picture, obviously the competition comes into the picture. So when competition
comes into the picture, you have to see, do we have the scale to assess a solution from the lens
of public sector delivery, from the lens of bringing an innovative solution to India’s market? So those are the things I think even the CAG has already laid out in their
sandbox approach through a framework.
As of now it is at a very basic level because it really put right keywords to add things because
since when we were discussing in our first question when we see that what kind of jargons are jargons?
Now are they jargons anymore or are they like common terms which are big? Exactly, exactly. And now
The gag has put the right words into the right picture, I would say, as simple as that is. And I would
suggest everyone can have a look at that. And they do talk about how we can assure it that AI is
adopted in a safe manner and how
assurance brings the safety of reaping the fruit of artificial intelligence solutions at the end of the day.
Aashka Patel (48:32)
Hmm. Yeah, makes sense. Makes sense. So if someone watching this wants to get involved in policy or governance or the amazing work that you have been doing and contribute to the country’s AI future, what is the best way for them to start?
Parishrut (48:49)
I believe that when I got into the picture it was somewhat by choice but by
the time frame I was like banking on artificial intelligence governance through some podcast I used to listen to with
it’s
With this essence or with this curiosity that maybe I will be able to like, know predict it Not make it easy. Let’s not talk about like, you Yeah, maybe I will be able to make a difference or not. It’s that the parameter for that is
Aashka Patel (49:16)
make a difference okay okay
Parishrut (49:29)
my work is having an impact of a certain sect of people not all together at large right
the Indian Hoi Palloy but for me if I would say I was really
going through nook and corners, through organizations, using the help of generative AI to really map out what the right voice is right now, back into zone 22, 21, 23. Like these three years I have really ventured into like
taking notes on my diary. I’m a bit old school with respect to that. Still very young.
Aashka Patel (50:02)
Yeah, Even
Sam Altman takes notes on a diary.
Parishrut (50:05)
The difference I can see right now is that there are right courses out there. There are
right platforms out there where you can learn about at free. No one is charging you. But obviously when economics comes into the picture, obviously you have to get accredited at the right place with the right bodies. And that’s
what makes all the difference.
And now if I have say
When I started, I used to brainstorm with AI. You act like an AI governance prof who
has been teaching for the last two years and quiz me.
And when Notebook LM came into the picture and it solved it for me, because now you can have quiz cards and practice your,
jog your memory and your understanding of a topic.
But apart
from that, I would say now, again, the Europeans have done a tremendous work in this. now they have, they will be introducing, it has already been passed, AI competency, like, know,
standard. So through that standard, you would know what
are the attributes or characteristics of someone who is in the field of AI. From the angle of
ensuring
artificial intelligence from the angle of ensuring artificial intelligence governance level technical level implementation level right and
apart from that i would say when these kind of instruments comes into the picture
different bodies will adopt these parameters and build on to something
whether it is a sandbox, whether it is a course and curriculum, whether it is
roundtable discussions, whether
it is further distributed materials for the public access. So now that has increased. So I would suggest that
for someone who is entering into this landscape, you click
the sector. Don’t leave the sector you are already in. Because it doesn’t make sense. You have to understand
from the angle that if
someone who has been working in climate change for a long long time they
can predict with respect to AI assurance what would be the difficulties just putting it out there so when artificial
intelligence is being diffused with
their respective sector I would say then they can they will be the most
informed human beings in a room to give their inputs and that’s what we need at the end of the day because
humans are in the loop to save us from artificial intelligence right so
Aashka Patel (52:51)
Yeah
Parishrut (52:55)
pick your sector, your niche, pick your competency wherever
you’re comfortable if you’re comfortable in creatives if you’re comfortable in
scientific exploration or from the angle of like different layers of artificial intelligence you can sense it and you can make sense through the academic knowledge you have. I would
Aashka Patel (53:15)
Mmm.
Parishrut (53:16)
say that go in that direction instead of coming
into AI governance outrightly because
a technical person can also comes to the AI governance. It’s
something a gem of the realm of a policy domain.
At the end of the day, all the domains which are functional
to keep a civilization working, to understand what is out there, what the humans have invented or what
humans have curated or created for the betterment of the society. So all the segments
of the society comes into the picture and the interdisciplinary approach has to be applied in this. And
we need experts of artificial intelligence in each and every field.
Aashka Patel (54:00)
Yeah, yeah, makes sense. So just to sum up, like, are there any policy or governance fellowships or courses that you would recommend to our listeners currently?
Parishrut (54:14)
I would say as of now we at Govern AI we have been doing in-person sessions and
Aashka Patel (54:19)
Okay.
Parishrut (54:22)
few online sessions as well but if you would want more structured obviously we can create it and we have been
creating for public sector officials and now
Aashka Patel (54:32)
Yes.
Parishrut (54:34)
for other sectors as well and apart from that
One which has helped me in my journey is of
Aashka Patel (54:42)
you
Parishrut (54:43)
BlueDot Impact It has really helped me to put
Aashka Patel (54:45)
Okay.
Parishrut (54:47)
certain impact certain impact
and through grounded intelligence because their syllabus is available because if I remember that
my fellowship with Equinia Institute where I was
Aashka Patel (55:00)
You
Parishrut (55:00)
taking I was facilitating this course to the emerging African leaders I had
my understanding of artificial intelligence from a very academic lens because
that has been my topic of like my doctoral studies right
Aashka Patel (55:13)
Yeah
Parishrut (55:15)
But
to put it into perspective, how
impact, how stakeholders has to be engaged and what are the key aspects to keep in mind. I think you need to get into the domains of certain understandings of the culture, which we will be catering to. So we were
really able to put into picture what
we had at that time.
syllabus with respect to AI governance right and
then I was really really in I think it was 2024 if I’m
not wrong in around July and around that time very recently I would say that this
course has been started by AI safety network of Africa. And I was like, yeah, no, I did apply to that. And the Equini Institute was the institution who was looking after that.
Liu has been one of the key figures who has made it possible at that time. And
in my sessions to facilitate, we were really able to map what we will be talking to
the
upcoming experts of like or like upcoming professionals of AI governance and AI safety in Africa. So,
blue dot impact is something it’s accessible. Yeah, you can it’s accessible. It’s free. can
Aashka Patel (56:42)
it’s the starting point.
Parishrut (56:46)
apply for it like and with the right understanding of the application process and then the Center for AI and Digital Policy. Their
program, their AI Policy Clinics has been
one of the key
flagship programs I would say it has
produced a good amount of like
officials
in this ecosystem and it’s again it’s also free of cost yeah so
these are the the components I would suggest through this
that everyone should
wrap their heads around and now
the state governments are like the state governments I would say that the country at the international level different countries are different states have been introducing courses which are free but
at the starting point these two are the main pillars through which you have to
get an understanding then put other things into perspective
Aashka Patel (57:49)
Yeah, thank you so much for answering all our questions patiently and accurately and to the best of your abilities. Thank you so much for your time. Yeah.
Parishrut (57:56)
I hope I was
able to do justice to each and every question and to the listeners also.
Aashka Patel (58:01)
Yes.
Parishrut (58:04)
Otherwise you can just obviously ask the rest of the questions in the comments once it’s out there in the public domain. But I love the questions, the line of the
questions and with what rigor and intent you come for it and I do appreciate and love it. Thank you for having me.
Aashka Patel (58:22)
Thank you
so much and let me stop the recording.
Aashka Patel (58:30)
hello and welcome to On Air with Aashka
Hello.
Parishrut (58:33)
I think there is the moment you press record, so lag in the, but we
Aashka Patel (58:36)
The lag? Huh. Okay, okay,
Parishrut (58:40)
can accommodate to it.
Aashka Patel (58:42)
Is it better now?
Is it better now? hello, hello, hello. It’s fine, right?
Parishrut (58:47)
Yes, yeah.
Aashka Patel (58:48)
Okay, So, let’s start again.
