- 3 hours ago
As Artificial Intelligence accelerates across industries, trust, governance and global alignment are becoming just as important as innovation itself. In this episode, we explore how AI standards are shaping Malaysia’s journey toward becoming an AI Nation by 2030 with International Organization for Standardization (ISO)'s AI Policy Lead, Cindy Parokkil, and National AI Office (NAIO) Policy Manager, Mr Darmain Segaran.
Category
🗞
NewsTranscript
00:07Good morning and welcome to Niagawani and today we're looking at one of the
00:11one of the most important questions that shaping the global economy which is how
00:16we build trust in artificial intelligence or AI and as we see that AI
00:21accelerates across sectors Malaysia is charting clear path towards becoming an
00:26AI nation by 2030 and central to the journey is the role of standards at the
00:32bridge between innovation and also responsible governance so to help us
00:36and unpack this we're joined by two experts at the heart of the global and
00:40also national AI governance ecosystem from the International Organization for
00:44standardization ISO Miss Cindy Perakil the AI policy lead and on her right is from
00:51Malaysia's own national AI office NIO Mr. Dharmin Seguran the policy manager thank
00:56you so much for being here Cindy and also Dharmin how are you thank you for
00:59having us amazing all right let's start the
01:02conversation at the global level before you localize it to Malaysia's path
01:06Cindy we see that the pace of AI deployment worldwide has accelerated
01:11dramatically but the trust and also accountability remain central concerns so
01:17from your global vantage point at ISO why have international standards become
01:22such an essential or important foundation for AI ecosystem today thank you for
01:27that question and I think at the global level since around 2018 the
01:32conversation very much has been around high-level principles what are the key
01:37principles that demonstrate that AI is governed responsibly so principles
01:43like transparency safety human oversight inclusivity so these are all
01:49principles that are being discussed and are critical for AI governance but then the
01:53question becomes how do you demonstrate that the AI system that you're deploying or
01:59designing is in line with these international high-level principles and that
02:05is where standards come into play because standards are that practical bridge
02:10between high-level principles and practical implementation because standards are
02:16essentially methodologies checklists frameworks that help organizations
02:21demonstrate that they are taking the right steps when designing and deploying AI in
02:29line with the principles that they have outlined all right thank you Cindy for that answer we've said
02:35we've seen the why globally but now we see we want to know on how at the global level as
02:42well so many
02:44governments right now are debating AI regulations also frameworks and also the
02:49AI specific legislation so with your experience at ISO how do international
02:54standards complement and also support and also guide at the same time the
02:58national regulation efforts especially by translating principles into
03:02implementable also readable practices of AI so governments have different tools
03:08available to help achieve policy objectives this can be regulation this can be
03:12legislation procurement policies etc good practice in terms of regulation is that
03:19regulation is not prescriptive not overly prescriptive it says what the outcome
03:24should be but it doesn't explain how to get there and that's where
03:29standards come into play because standards can provide the methodology the
03:34practical guidance in terms of how to achieve that objective so I think the key
03:38difference that we need to keep in mind here is standards are voluntary tools so
03:45regulation and legislation are mandatory tools and the magic happens with the
03:50interplay between the two regulation and legislation are tools that governments
03:55have that they can use to achieve various policy objectives but good practice in
04:00terms of regulation is that regulation is not overly prescriptive so the
04:05question then becomes how do we achieve these objectives that are described in the
04:11regulation or in the legislation and that's where standards come into play
04:14because standards essentially are checklists their processes their
04:18methodologies that can help demonstrate that a company is developing and
04:23deploying AI systems in line with good principles good practice etc and I think the
04:30the main point to keep in mind here is that standards by themselves are voluntary
04:36tools however regulators if they believe that these standards are good and they
04:43are developed in a system in a open consistent consensus-based manner they can
04:48point to these standards as a way to demonstrate compliance to regulation so for
04:54example in the information security space government can say because governments can indicate or make it a
05:03requirement that companies that wish to handle government contracts which deal
05:09with you know sensitive data they need to show that they comply with 27001 ISO IEC 27001
05:17and this is a way that there is an interplay between voluntary standards and
05:22mandatory tools thank you I'll get back to you on the standards that you mentioned
05:27earlier but Darmin building on what Cindy shared about global momentum we see that
05:33with a global frame that we want to bring on the skill of local or Malaysia's roadmap
05:39up towards 2030 so how does the MI sorry MY AI standards am I pronounce it correctly or my AI
05:47standards
05:48initiative can actually support that national vision towards 2030 thanks thanks for the question the the vision of the AI
05:56nation was set out in the national plan last year by the prime minister and I think all policies are
06:02going to be heading pointing in that common direction
06:04so the national action plan AI action plan that our office is working on and hopefully will be produced soon
06:12by the Ministry of Digital
06:13this points towards again developing an AI nation and a core theme of these policies and these drives is increased
06:21adoption
06:22and increased adoption across industry uptake by developers within within the spectrum of the ecosystem this requires a core fundamental
06:32of trust
06:33trust and I think that's where you'll see the pillar being drawn out across these policies the common thread that
06:39cuts across which is
06:40trust in these systems and if you want users to be comfortable and using AI systems to be comfortable in
06:46this existing in the ecosystem
06:48trust again is a core element and that's where this initiative has come about as Cindy explained earlier it's the
06:56link between these two elements
06:58between regulation and turning it into something that's operable on the ground
07:04and that's where standards play a role so the my AI standards initiative is focused on bringing awareness to standards
07:11and their potential use in many different ways not necessarily just for compliance but also for policy building
07:17framework building within organizations and even by policy makers themselves
07:21and in Malaysia that trust that translation shows up in the sequence of policy makers have chosen
07:27and to which Malaysia has taken a deliberate step-by-step approach which is principles first and then standards
07:35followed by regulation also legislation so what's the strategic logic for sequencing governance this way in a fast-moving
07:43technology space especially in AI so I think the the sequence the approach is one way of looking at it
07:50but with the speed of how
07:52AI is being deployed what you'll tend to see it's also a sort of a parallel approach almost a layered
07:57approach now I think Cindy
07:59mentioned earlier that principles have been the global trend and we saw that since 2018 till about now where principles
08:06sit at the core of most
08:08frameworks around the world and the need for translation of those principles
08:13is where standards come in now moving beyond that is where in a certain high risk uh deployment sectors
08:20uh certain use cases that pose that excessive risk you'll see more regulation and legislation
08:26that might need to tie this together the governance frameworks together now all this is happening at
08:32common pace actually so in Malaysia uh the minister announced development of the AI governance bill
08:37and that's being developed but in the meantime the ecosystem needs to be ready because in a situation where legislation
08:45is deployed and the whole ecosystem
08:47the industry is not ready for this large shift you'll see a failure of that operation of that legislation and
08:53so what this
08:53initiative is in to do is actually begin that process of getting comfortable with uh understanding these principles how would
09:01they work in
09:01practice and working out the kinks before regulation really kicks in to refine makes sense um now we want to
09:09get specific about the tools organized organizations can actually adopt which you already mentioned earlier Cindy um
09:15two anchors keep being referenced which is the ISO 42001 the AI management systems and also the ISO 23894 the
09:26AI risk management
09:27so how do these two standards actually help organizations operationalize responsible AI on the ground uh just
09:34beyond policy statements so i think standards at the end of the day are very practical tools that
09:41companies can use when they think about deploying AI in their organization or designing AI um in that in that
09:49um for a matter of fact so it it provides this framework to them to show and demonstrate that they
09:57are
09:57developing AI in a responsible manner so if you take ISO IEC 42001 it it poses very pertinent questions that
10:08organizations should think about when they put AI into an organization for example do you understand the context in
10:15which AI is being deployed because the AI system will interact with you know the organizational context
10:22and the larger context is that has that been thought about is there leadership commitment what are the
10:28controls that are being put in place how are you coming up with a policy so it gives you that
10:33kind of
10:33framework and the beauty of standards is that it can be used by organizations of any size in any sector
10:42um so it's this versatile agile tool that helps it provides a starting point um to organizations when
10:51they are deploying AI in a responsible manner um there's another standard that has come out ISO IEC 42005
10:59on impact assessment so this basically helps an organization in thinking about and showing and
11:08demonstrating an impact assessment of AI so rather than coming up with your own uh you know way of
11:14demonstrating um that you have done an impact assessment look if there is an international good
11:20practice that is out there that you can use and follow to um to demonstrate that you are taking the
11:27right considerations on board so basically talking about standards so much to be considered so much to be
11:33thought and it's not uh written in silos they're actually forged by coalitions from contributions
11:39across governments uh industry players and also researchers and many more uh so why is that multi
11:45stakeholder approach essential for AI especially for something as complex and also fast moving um AI
11:52governance and how does it actually strengthen both adoption also accountability at the same time
11:57thank you for that question so i the first point i'd like to say is standards are a product of
12:04multi
12:04stakeholder collaboration be it for AI or for any other topic area when we develop standards um from an
12:12ISO perspective from an ISO IEC perspective on this um on this matter we make sure that all voices are
12:19heard
12:19it is very important because standards are a product of consensus and it is very important when we
12:26develop good practice that we have all the different stakeholders um that are impacted by the topic
12:34matter um are around the table so this is industry government academia civil society researchers so we
12:42really try to bring all these experts together to develop international good practice because it's
12:48important to hear all their voices in an area like AI or in technology the issue at our disposal
12:55is not just a technical one it is a socio-technical one meaning that people are being impacted so it
13:03is
13:03really important to have these people around the table who are being impacted by AI and that's why
13:11multi-stakeholder collaboration is a key principle that we follow in standards development
13:19and that brings to my next question to darmin um when malaysian companies look for practical help
13:25where do they start because darmin um for example when malaysian organizations like SMEs for example visit
13:32the my AI standards platform for the first time what practical support or clarity do they gain and how does
13:39it
13:39help them navigate the often overwhelming landscape of AI standards and also frameworks i just mentioned earlier
13:44yes i think overwhelming might be the right word to use uh the the my AI standards initiative has an
13:52overall high-arching aim which is to increase awareness and use of standards uh by various segments of
13:59stakeholders so that's policy makers industry academia even and to achieve that one of the starting
14:04points is the portal that is said to be launched uh this portal the immediate focus is navigating these many
14:11standards that exist and it starts with the ISO uh standards that exist which is a number in about 80
14:18standards in total now when i say navigating what i refer to is not all standards apply to every single
14:25sector industry so members of certain segments of all sectors will be able to uh filter through to find
14:31relevant sectors for their own use now that would be a starting point now moving from there there will be
14:39uh increased levels of uh awareness programs around this there will be uh use cases that might be deployed
14:45and the portal is set to go in that manner but the broader initiative is to increase the ability to
14:51use
14:51these standards internal policies for example if an organization wants to build internal policies uh don't
14:58start from zero refer to the standard to see what frameworks are available there and build from there so
15:03when you do that you find yourself being in alignment not just with local best practices but
15:08international and global best practices which automatically makes you interoperable so if
15:13you're reaching out to wider markets and those standards are in play there then you're already
15:18interoperable and let's translate that approach or standards into outcomes for malaysia's key sectors
15:24um where are you seeing the earliest maybe most tangible benefits of standards aligned ai in malaysia
15:30whether in safety interoperability uh auditability or time to deployment across healthcare across sectors
15:37like manufacturing also smart cities maybe are there any examples that you want to share or patterns that
15:42you can actually share okay so i think the sectors that you mentioned uh earlier are those critical
15:49immediate sectors that will have immediate reference points uh in the portal when it's launched but there will
15:54also be a segment of cross cutting standards so it cuts across all uh sectors uh all of these point
16:01towards an intention for industry to become digitally resilient and what that means is in the deployment of ai
16:08systems where there is a risk uh of failure of these ai systems that may have high impact to uh
16:17certain
16:17stakeholders or to users this is where the standards will aim to actually make those organizations resilient
16:24and be able to continue to operate or to have fallback measures right within uh so that the impact is
16:31mitigated and that operations can continue so that's one of the uh key aspects for these uh identifying of
16:37these critical sectors now uh while these might be the first few to be uh filtered through on the portal
16:43there is an intention to expand this to other sectors and build and continue to build this uh portal in
16:49terms of uh content and availability as well as uh the manner of navigating these different standards
16:55and this is for sunday um let's relate back to malaysia where we see that countries like
17:00malaysia are rapidly developing their ai ecosystems so um how can actually participation in international
17:08standards uh help emerging digital economies accelerate responsible ai adoption
17:13i think this goes back to the point that darmein made earlier about trust right you in simple terms
17:22you trust things that you have a you've had a hand in making so by participating in international
17:30standardization you're essentially co-creating this international deliverable so i'll just say a few words
17:36in terms of how iso operates right so we at we iso we're essentially a network of 175 national standards
17:46bodies one member per country in malaysia that member is called jsm okay so jsm has a very important
17:53role to play in malaysia in that they convene all the stakeholders that i mentioned earlier government
18:00academia um industry etc to collect input and provide that input back to the international
18:08system um based on what the needs of the country is so that process is very important in creating
18:16these standards so by participating in standards you're essentially co-creating these standards
18:21meaning that will be easier to adopt them because they're also reflecting the needs of malaysia so
18:29that's why participating in international standardization and having your voice heard is very important
18:34to increase the uptake of these standards and thereby deploy ai responsibly in your economy
18:40and that plays directly into malaysia's partnerships with iso so um malaysia we know that the true
18:47programs such as the ai policy capacity building journey and also we have global dialogues between
18:52iso so how important are these international partnerships for malaysia's ai development moving
18:58forward i think it's absolutely critical to continue these dialogues and uh the important question there is
19:04the spaces in which these dialogues occur uh they vary so depending on whether you know uh it's a global
19:10dialogue between countries but also as you mentioned the capacity building programs which are much
19:15more workshop style intensive uh programs uh like the one that iso ran uh which can result in
19:21impressive outputs uh such as this and we see the important spaces here that i spoke about is
19:27spaces for interaction and learning uh between our country and other countries as well and not just
19:33on the success points but also the failure points and uh in the room when you're having these open
19:37discussions is where we learn and we also understand how we can further collaborate i understand that
19:42this capacity building program is going to continue through due to its successes and malaysia is keenly
19:47going to be involved in that moving forward and uh before we wrap up maybe we want to give on
19:53the first step towards the companies how can actually act on ai deployment towards the organization so
20:00maybe we can start with cnd first for organizations they're just starting that a journey what is the
20:05single most important first step that they can actually take to align with trusted ai standards
20:12because for some companies they can find the ecosystem confusing so what can you suggest them
20:17so rather than reinventing the wheel based on the objectives of the company i would suggest as a
20:24first step visit the my ai standards portal to see what standards are out there and try to map out
20:31which of those standards can help you as the organization achieve the objectives that you
20:36have laid out how about you darmin final world to malaysian business watching right now um they are
20:43experimenting right now with ai but not yet thinking about governance so what's your practical message
20:47to them as they scale their ai adoption industry is definitely uh growing on the idea of governance that's
20:53necessary for ai and those that are not thinking about it i think it's the message is to begin thinking
20:59about it and maybe in the context of standards as a starting point and the reason is because the
21:03risk that is being posed by ai sometimes spills beyond your own organization so you have to think
21:09a bit wider than your own organization your own industry impact wise it can be significant so yes
21:14echoing what cindy has said starting potentially with visiting the portal and understanding and then
21:20building on that to governance frameworks within your organization aligning with the national strategy
21:25all right thank you so much to cindy and darmin again for both of your insights so what's clear
21:31right now is ai leadership isn't just about developing the advanced technologies but it's about building
21:36trust and malaysia is positioning itself to innovate responsibly and confidently on the global stage so
21:42thank you so much for watching you can find this whole discussion on all of our social media platforms
21:46and that includes astroone.com see you next time
Comments