- 1 year ago
- #agendaawani
As AI reshapes our world, where does ASEAN stand in the global safety debate? Hear from Dr. Jun-e Tan, Senior Research Associate at Khazanah Research Institute, on how the region can position itself in this crucial conversation. #AgendaAWANI tonight at 830pm.
Category
🗞
NewsTranscript
00:00Asalaamu alaikum and a very good evening I'm Travis Anjohari here at the ASEAN AI Malaysia
00:23Summit 2025 now there's a lot of conversation about AI and we're among the
00:30conversation is of course the governance and innovation but amidst of that we
00:35also were triggered to talk about these AI safety so to talk more about you have
00:41safety here with me is Dr. Juni Tan she's a senior research associate at Kazana
00:46her research thank you for you know allowing yourself to be free and today
00:53today is a very busy day and I think this is among the very first and big yet
01:00important event when it comes to AI in Malaysia even though AI itself is still
01:09very very new but there's a lot of progress especially there's a lot of
01:14commitment from the government to to push through this innovation and there's a
01:19lot of talk about innovation of course there's a lot of governance but there's
01:22not much talk about the AI safety and this is where you're coming in right so
01:26can you talk about okay can you tell us a bit you know walk us through what does
01:30AI safety mean vis-a-vis AI regulation right so AI safety is basically a way of
01:38considering risks that AI whether it's safe or not if it's built properly or not
01:43so that when you actually deploy it and it's out in the open the technology is
01:49safe to use right so we are talking mainly about say the reliability of it
01:54whether it might have unintended like design flaws or unintended consequences and
02:01it is a little bit different from say malicious use of AI which is like a more
02:06a security issue so AI safety is more risk of this this this type
02:12unintended yeah yeah and out of design like how can we make AI safer now this
02:21industry if we can say I as an industry is still very very new especially here in
02:29Malaysia how do you bring this conversation
02:35into the domain how does the innovators for instance are taking it so if we
02:45bring it back to like Southeast Asia because this is an ASEAN AI summit so I
02:50think that our priorities might be slightly different from outside of the
02:55region because AI safety conversations outside of the region focus a lot on
03:00like the future harms of AI for instance chemical biological radiological and
03:06nuclear CBRN risks or you know risk of artificial general intelligence or
03:13super intelligence you know but on our side we're a little bit more pragmatic
03:18and we are thinking a little bit more about immediate harms that AI might bring to
03:23the region and how we will practically try to mitigate these harms so there are the
03:29differences when it comes to the discourse of AI safety this region ASEAN
03:34compared to maybe a more advanced economies yes might sound a bit more
03:39surreal kind of in the movie discussion hours are more practical as you say can you
03:47give us a bit of example so practical concerns about AI safety would include like
03:54online safety issues so that would look at deepfakes for instance or scams AI
04:00enabled scams or it could also look at whether we have enough of data to train
04:07proper you know models or not because if you don't have good data then you'll end up
04:12with models that are skewed and that they won't work as well right so algorithmic
04:18bias we are a very diverse region with a lot of existing fault lines yeah so how
04:24how we do we make sure that AI will not exacerbate certain like relations within
04:30society so all of these are actually the AI safety issues that we in the region have
04:35in mind and also it reflects our kind of status as users more than producers so
04:44outside the region they're talking more about frontier AI model risks because
04:48that's a little bit more when they're developing what are the things that they
04:53should take care of whereas on our side it's when we are users how do we make
04:58sure that AI is used properly it's designed to be safe to be used right such
05:03concerns I believe there's a lot of also there's also a lot of this when it
05:10comes to the conversation like this conversation of AI safety there's a lot
05:14of also debate when it comes to especially guidelines for instance does we have a robust
05:22guidelines yet I know at ASEAN level there's some sort of guidelines as a
05:31you know like that can help us to build us in a safe environment we also I think
05:40we're also talking about Malaysia or national level guidelines where are we
05:45now when it comes to that guidelines so we do have guidelines at the Malaysian
05:50level and we also have guidelines at the ASEAN level and so if we look at the
05:55ASEAN AI ASEAN guide on AI governance and ethics I find it to be a rather
06:02practical I will keep saying practical meaning it's not too restrictive meaning
06:08that it's actually useful all right right so if you look at the guidelines the
06:13ASEAN guide it's it's kind of a bit of a quick crash course on what you should
06:19think about when you want to implement AI at an organizational level so what is
06:24defined as AI how how should you change your organization to to sort of incorporate
06:30AI what are the principles that should you know underline AI use all of these are
06:36addressed and they also root it down to the AI system lifecycle and also use cases
06:43which is important because you do want to demystify AI you want to know how
06:49it works and where it can be applied on so that I do appreciate from the ASEAN guide
06:56and also lastly on the AI risk impact assessment template that they have at the
07:02very end I think that that's also something that's quite useful so for within
07:07ASEAN if any organization wants to apply AI they can actually use it we are going to our
07:14first break and then when we come back we'll talk about AI safety network there
07:18there has been conversation among our players in the region of course regulators
07:23in this region about this network the AI safety network and how can this be a
07:31stimulus to bring this industry forward
07:34and I'll see you in a bit
07:38welcome back to Agenda Awani and we're recording this episode in the Junction of the ASEAN AI
07:44Malaysia Summit 2025 we're still with Dr Jun we are talking about the AI safety we're talking about F-70 in Malaysia as well as ASEAN region of course
07:53this is an ASEAN event event there has been conversation about the ASEAN AI safety network
08:05how's it been going so far and what is your expectation what can we do with the ASEAN AI safety network
08:11right so the ASEAN AI safety network is still being discussed it's at a feasibility study stage but hopefully it will go through and it will become like a focal point or a network hub for ASEAN member states national AI focal point
08:30It's at the feasibility study stage, but hopefully it will go through and it will become like a focal point or a network hub for ASEAN member states' national AI focal points.
08:41So that is the hope that it will be that network.
08:45And I see its usefulness in three different angles.
08:49One is that it should, I hope that it will be a key conduit on building AI readiness because we have all of these different countries at different levels of maturity.
09:00So you really do need to have capacity building.
09:03And then secondly, within the network, you'll have a better sharing of information, research, or communication in a more efficient way, right?
09:14You have something happens in one country and then it spreads to other countries because there is a network of this information of this new threat, for instance, which is where things like incident reporting will become quite useful, right?
09:29So something happens in, say, Myanmar, and then so all the countries are notified of it.
09:35And of course, but a lot of work has to be done on how to define an incident, you know, what are our priorities and all that.
09:42And compatibility across the region.
09:45Exactly.
09:45Exactly.
09:45Like to arrive at that with consensus.
09:48And the third thing that I see it to be useful is that it should actually foster international collaboration with the other AI safety institutes that are all over the world.
10:00Like there's one in Japan, there's one in Korea, US, China.
10:04They all have like AI safety institutes.
10:07So if we are, we have like a network, this will be the focal point where collaboration can happen with the other institutes.
10:15Who sets the standard down?
10:18So that's an interesting question because the standards conversation usually is ISO.
10:26So it's a multilateral kind of, or a multi-stakeholder kind of platform where industry players or civil society or governments can get to talk about it together.
10:38But I also think that the standards conversation is starting to get a little bit fragmented because of different priorities in different countries.
10:48So hopefully we will actually keep it together because the standards are useful mainly when everyone agrees to them.
10:57Are we expecting different kind of standards that we will apply here compared to, let's say, Europe, Japan or the US?
11:06So I think that right now, for instance, Malaysia is a participating member of ISO.
11:10So we actually bring the standards of ISO down to the Malaysian standards level.
11:19So it is in our interest to have not that many standards because then when you want to work on compliance.
11:25There will be an issue of compliance.
11:26Exactly.
11:27So that will be cost to businesses.
11:29So it is in our interest to actually make sure that our views and our voices are actually input to the standards setting committees
11:38and development organizations, you know, at the international level so that we don't have, like, as you say, many different standards.
11:46That's right.
11:46Oh, good.
11:47So now, coming back to the more practical conversation, rather, I guess, there's a lot.
11:59When it comes to technology adoption, there's a lot of conversation that needs to be made when it comes to trust.
12:05We had this conversation many years ago, and it just ended recently about the digital and public trust.
12:14And now we have to talk about AI technology and public trust.
12:18Yes.
12:19How can AI safety help to navigate this conversation to a more progressive and positive tone?
12:27Right.
12:28So when we talk about trust, I think we'll have to look at it from the adoption and the governance point of view.
12:35Right.
12:35Because without adoption, then there's no point in governing, but governance will actually spur adoption, right?
12:42So for this, we really need to make sure that there is trust in AI, because if there is some scandal,
12:50there's some big thing happening that indicates that AI that we use is not safe, and it has happened before, right?
12:56Right.
12:57Then public trust towards the technologies will be affected, and adoption will be set back.
13:03So governance, I see hand-in-hand with adoption, because it's a more sustainable way.
13:11So if you unleash innovation without safeguards, at some point, you're going to have something bad happen,
13:17and then the public will lose their trust, and then you have to start to build that trust again,
13:22and that that's just a setback, yeah.
13:24So you're saying that even though while we are pushing positive innovation and adoption,
13:31at the same time, regulation has to be robust to support the growth.
13:35Yeah.
13:37Where are we now?
13:39So when we talk about regulation, it's important to also note that there is non-legal types of regulation,
13:47and also legal types, right?
13:49So there should be an array of these different types.
13:54So right now, the sense that I get from Southeast Asia from the ground is that we don't want too much of a legal type of framework,
14:02because we're still kind of waiting and seeing, and the adoption level that we have does not warrant high regulation as yet.
14:11So the sense that I get is that we have to be agile, and we have to actually build the processes more than regulating AI directly.
14:23So like for instance, if you have an AI safety institute, so that institute will be doing research,
14:29it will be doing tracking and all of that, and that will actually enable you to, at some point,
14:35if you want to regulate, you can do it with information, right?
14:39You can do it without being blind about it.
14:42Right, right.
14:42So the way I see it is that right now is a really good time to build frameworks and processes and bodies, for instance,
14:51so that we can stay on top of where AI is going, and what are the possible risks, safety, or security risks that we see.
15:00Right, okay, I think we're heading on the right trajectory when it comes to this conversation.
15:07We have to take another short break, and when we come back, we'll talk about Malaysia as a leader in ASEAN when it comes to AI.
15:18When it comes to AI, what are the opportunities, and how can we foster a stronger collaboration with our partners in ASEAN?
15:25Right, yep.
15:26Right, and I'll see you soon.
15:27We'll go away.
15:28Okay.
15:29Okay.
15:43Welcome back to Agenda Awani.
15:44I'm Shafazan Johari, still with Dr. Jun Yee.
15:48We're recording this episode in conjunction of the ASEAN AI Malaysia Summit 2025.
15:54Now, ASEAN and Malaysia and AI, these are the three keywords that we've been talking about since the last two blocks.
16:01Coming to our third block, the final block, I would like to ask you about our positioning as a leader in Malaysia.
16:09Does we have any advantage to carve ourselves as a leader in AI or any area, especially when it comes to policy in the region, of course?
16:20Right.
16:21I want to start from the national level.
16:23Yeah.
16:24So, from the national level, I think there's a fair bit of political will when it comes to AI policy and spearheading AI for the country's benefit.
16:32So, we are actually not only talking about it, we're also putting resources in to push.
16:38Right.
16:39So, like for instance, we gave resources to set up a national AI office, which has been really, really useful.
16:45Because previously, there was a lot of fragmentation in what's going on in AI, but now we actually have like a focal point.
16:52Right.
16:53So, the focal point is really important because it's able to then bring people together from different interest areas.
16:59Say, AI safety or AI talent, you know, or AI ethics, you know, different types of working groups have come together.
17:09And the other part of it is to actually bring AI to the sectoral level.
17:13Right.
17:14So, how does AI look when you look at finance, agriculture, health, you know.
17:20So, a focal point like NIO, which is the national AI office, is very helpful to make it like efficient, you know.
17:27Right.
17:28It's a little bit like the ASEAN AI safety network thing that I was talking about.
17:32Which brings me to the regional level.
17:34So, the ASEAN AI safety network, the proposed network, is actually a Malaysian initiative.
17:41And I think that that is really important.
17:45And our role in this is because we're not super in front in the race, but we're not super behind either.
17:53So, we can sort of see from different perspectives.
17:55Right.
17:56How the technology is going very fast and what are lacking when it comes to, say, the safety conversation or the governance conversation.
18:05So, the ASEAN AI safety network will be the hub to actually uplift the entire ASEAN, hopefully.
18:12So, that will, yeah, that is how I think Malaysia will be able to show some leadership and some lessons learned to other ASEAN member states.
18:23And at the same time, we're also seeing a lot of players in the industry.
18:29I've seen a lot of, actually, businessmen and entrepreneurs have shifted from more traditional business into AI.
18:38How are we harmonizing this conversation, especially when it comes to governance, regulation and safety?
18:47Right.
18:48So, on that, I think we had talked about standards.
18:52Yeah.
18:53We had talked about the importance of having harmonized standards.
18:55Yeah.
18:56So, that is something that we need to have.
19:00And I think it's good that entrepreneurs are looking into this and startups and things like that as well.
19:07But I guess the difficulty is looking for use cases that make sense.
19:12And that is where a lot of information sharing and a lot of learning from other application areas will actually help uplift the industry.
19:24Meaning you still have to work in a team.
19:26That is where I usually stand.
19:28So, that's why events like this are so important.
19:32Yes.
19:33Because you are capturing everyone from different industries.
19:37Yes.
19:38Even from media, like us in Eswar Wani, to come and have this conversation.
19:42Yes.
19:43Our application, our AI application might be different.
19:46For instance, you're in research and I'm in media.
19:48Yeah.
19:49And others in, let's say, in health tech.
19:50Yeah.
19:51Our application might be different.
19:52How we use energetic may be different.
19:55Yes.
19:56Right.
19:57But for the sake of AI safety and for the sake of the public trust, we have to work together.
20:02Yeah.
20:03I think so.
20:04I think we actually need to have a lot of conversations.
20:07Because ultimately, we'll have different visions of what we want to see as an AI future.
20:13Right.
20:14We would have different visions and we would also have different associated risks.
20:18And especially when you go down to the sectoral level, you're an expert within the media side.
20:23Yeah.
20:24And then you have other people from say agriculture or say finance, they will be experts in their own domains.
20:30Right.
20:31And so, bringing that domain expertise in will help us understand much more.
20:36Right.
20:37So, AI is going to change things and how to actually adapt to the changes.
20:44Yeah.
20:45And then that's also when we see the gaps in the safety.
20:48Yeah.
20:49Because I might not know why or rather how a banker, the banking industry, the banking industry
20:55and finance, which we are using AI.
20:56Yeah.
20:57And how kind of risks they are taking when it comes to AI.
21:02Yeah.
21:03Now, so, in an event like this as well, you also have policy makers.
21:08Not just innovators.
21:09Yeah.
21:10And entrepreneurs.
21:11Pushing the agenda forward for the regional AI safety, this is the first day.
21:16We've just been here for the event just started for the last couple of hours.
21:21Yeah.
21:22But what kind of expectation and hope you have for regional cooperation like this?
21:28Right.
21:29Okay.
21:30And how are we going to ensure that this kind of cooperation are not just the side or not
21:38just, you know, it can and it should and it has to be a real cooperation.
21:43Right.
21:44To push things forward.
21:46So, the way I see it is the ASEAN AI Summit Malaysia is kind of like, it's not a one-off.
21:52Right.
21:53It's one conversation out of many to come, hopefully.
21:56And why is that important?
21:58It's important because there are also a lot of global summits happening.
22:01For instance, the AI Action Summit that happened in France earlier this year.
22:05Right.
22:06And the AI Impact Summit that's happening in India soon, that a lot of these conversations
22:11will be happening at the global level.
22:14Right.
22:15So, for us, it's important, I think, to have like a pre-conversation so that we can actually
22:20strategize on what are our common ASEAN interests, what are the things that we would want to negotiate
22:28for and push for and how do we connect with our peers in other regions to advocate for, you know,
22:36global AI governance that benefits us, right, as the region.
22:41So, as the ASEAN, you know, Southeast Asian region.
22:44So, this would just be like a great platform that I hope that is not, you know, for one time.
22:53It will happen yearly.
22:54And those conversations will then feed into other international fora that then will be able
23:01to push for, you know, better AI, safer AI for all.
23:05And this is also the role of researchers like you to actually listen to all of the conversation
23:12and identify those challenges and gaps.
23:15We try our best.
23:17I recommend a way forward.
23:19Thank you so much, Dr. Juni Tan.
23:21This is all the time we have for this episode.
23:23Thank you for sharing your insights on AI safety because this is very important.
23:28We need to talk about this because this relates to the public trust.
23:34And without public trust, it's very hard for us to push forward when it comes to adoption.
23:40Thank you so much, Dr. Yi Tan.
23:42This is Juni Tan.
23:43This is Agenda Awani.
23:45We recorded this episode in conjunction of the ASEAN AI Malaysian Summit in 2025.
23:51And that's all for this episode.
23:55I'll see you in the next episode.
23:57I'll see you in the next episode.
23:58Thank you so much.
23:59Bye.
Comments