AI and Journalism — a convo with Karol Ilagan

It’s still the early days of AI in the Philippines, but journalist and UP professor Karol Ilagan describes how AI tools can already impact journalism, from the basic task of transcribing interviews to combing dense COA reports in search of patterns and red flags.

HOWIE: Magandang araw, podmates! Si Howie Severino muli na nagpapaalala na nakakatalino ang mahabang attention span. Ang guest natin ngayon ay isang journalist, professor at researcher na naka-focus ngayon sa Artificial Intelligence o AI at ang mga epekto nito sa media at sa ating lahat. Si Prof. Karol Ilagan na nagtuturo ng journalism sa UP Diliman. At matagal ko ng kasama sa PCIJ kung saan ako'y naglilingkod bilang chair ng Board of Editors. Magandang araw sa iyo, Karol. And welcome to my podcast.

KAROL: Magandang araw, Howie. Maraming salamat sa imbitasyon.

HOWIE: Karol, I was in the audience last week sa iyong presentation sa Third National Conference on Investigative Journalism. Ang pamagat ng iyong talk noon ay Co-opting AI o Artificial Intelligence, Investigating AI. So please tell us about this research project. Maraming namangha roon sa audience. So what do you mean first by Co-opting AI?

KAROL: This research is for a book. Ang tawag du'n sa libro ay Future of Media in Asia. So it's a book series. So in particular, may pagtutok sa media. Titingnan ano ba ang magiging future natin.

Well, maybe for a few years already na pinag-uusapan 'yung AI. So naisip ko na tingnan natin kung kumusta na kaya 'yung uptake ng AI sa Pilipinas at sa iba't ibang bansa in Southeast Asia. Although kailangan kong banggitin, ano, Sir Howie, you know naman na the research is yet to be published. So I might not be able to provide details. Pero at least 'yung broad strokes, maibabahagi ko naman po.

HOWIE: So okay. So you're researching the uptake. In other words, the adoption, kung paano siya ginagamit, anong extent ng paggamit. I presume hindi masyado pang ginagamit ang AI sa Pilipinas. But may mga uses na. So paano ba ginagamit sa kasulukuyan ang artificial intelligence sa ating bansa?

KAROL: So hindi po kayo nagkakamali roon. Medyo low pa talaga 'yung adoption niya or 'yung paggamit ng AI sa atin. Kung nagagamit man siya, more on nakuwento ko ito na specific to a task. So for example, magta-tanscribe tayo. So 'yan 'yung mga common na ginagamit natin 'yung mga AI tool. Or 'yun nga, ang mas common ngayon na pinag-uusapan ay 'yung chatGPT or 'yung generative AI.

So mababa pa 'yung adoption natin. Kaya what got me interested dito ay parang 'yung ating kasing conversation on AI or the narrative or 'yung discourse ay parang AI, ay chatGPT. So parang 'yun 'yung usapan natin when 'pag titingnan natin kung paano siya ginagamit in many other countries, ang laking bagay noong AI in particular to investigative reporting, for instance. So kumbaga, marami tayong examples ngayon of stories that would not have been possible kung hindi gumamit ng AI. And I'm not speaking of generative.

HOWIE: Excuse me, Karol. Before we proceed, I will ask you about how it's being used sa investigative journalism in other countries especially. Kasi hindi pa masyadong ginagamit dito sa atin sa IJ. But nabanggit mo na rin 'yung chatGPT. That is pretty widely used now in the Philippines. Pero maaaring may mga listener tayo na hindi pa nakakaalam kung ano 'yan.

But basically, chatGPT is just one tool that you can ask on the internet. It's artificial intelligence-driven, na makasasagot ng maraming tanong. And it can actually perform tasks for you. You can give it an assignment like "Write me a speech about investigative journalism in the Philippines." Gagawa siya ng speech.

I mean, hindi naman guaranteed na 'yun ang gusto mong sabihin or maganda 'yung ibibigay sa 'yo. But mayroon siyang maibibigay, right? So of course, chatGPT is just one version nitong AI tool na ito. I know my son uses BARD, 'no? And then other people use other tools, et cetera.

KAROL: 'Yung chatGPT, it's an online tool. 'Pag pinag-uusapan namin ito ni Jaemark, siyempre may iba't ibang application. Pero ang pinaka mas nagagamit siya ngayon ay think of it as parang maggo-Google pero contextual 'yung ibibigay niya sa iyo na information.

So pero dahil nga, nabanggit natin 'yung generative, nakaka-produce siya ng content. Kaya naging controversial siya because 'yun nga, nakaka-produce siya ng content na parang tulad natin when we produce content also.

HOWIE: At saka, 'yung malaking kaibahan nito sa mga nakasanayan natin na search engine is that it does not provide you with any links. It does not bring you anywhere else. Hindi tulad ng Google. Kapag mayroon kang nire-research, for example, if you want to search AI itself, Google will produce a whole list of links of websites where you can read about AI.

But AI tools themselves will not give you any links. They will just give you what it thinks you want to know. 'Di ba?

KAROL: Right, right.

HOWIE: And this has a big effect, actually, on 'yung tinatawag na search referrals that's a big source of traffic and eyeballs for a lot of websites, a lot of media pages, news sites. So ang laki ng impact nito in terms of 'yung internet models, business models ng mga web company, websites. As you mentioned, people are doing their search on AI tools, which will not bring you to another webpage unlike Google, 'no? So it's actually changing the way we're using the internet.

KAROL: Right. Isa 'yan sa malaking issue with regards to generative AI tools like chatGPT because 'yun, I think may ilang news outlets na parang pinapa-block nila para hindi magamit 'yung mga news report or information from their websites.

Although, I think, ang bagong development din ngayon, may mga iba naman na nagkakaroon ng licensing agreements na with, for example, chatGPT para magamit 'yung information sa kanilang website. So medyo, ano po ito, mabilis 'yung mga development. Obviously, there are a lot of considerations. We've talked about ethical considerations already. And obviously, 'yung impact rin on news outlets given na ang konteksto rin natin, of course, we're talking about disinformation as well. Parang even more so ngayon, 'di ba, mas kailangan natin, na 'yun nga, mas ma-access 'yung legitimate news sources.

So ayun, medyo halu-halo 'yung issues with regards to AI tools, Howie. Gusto ko rin sanang magkaroon ng kaunting shift doon sa ating usapan because obviously, granted 'yun, we are not discounting 'yung mga ethical consideration. And 'yun nga, impact on our jobs. Maaaring threat 'yun sa ating jobs. Not just in journalism but lahat yata anong sectors and fields of work puwede niyang maapektuhan.

Ang iniisip ko ay whether baka we're missing out din naman doon sa potential niya sa ibang bagay. Ang ating discussion ay medyo 'yun nga nakasentro doon.

And tama naman na dapat siyang pag-usapan. Pero 'yun, kaya siya Co-opting AI, babalikan pa 'yung tanong kanina, we might be missing out on other ways na para magamit 'yung AI to advance reporting or advance our work as journalists.

HOWIE: Okay. So, based on your research, what are we missing out on pagdating sa AI? I mean, ano ba 'yung ginagawa na sa ibang lugar, ibang organizations or countries na puwede na natin sanang gawin ngayon dito sa atin at makatutulong sa ating trabaho bilang journalist?

KAROL: Okay. Maraming mga nag-aaral tungkol sa AI and journalism. So isa na riyan ang Reuters Institute at saka 'yung isang organization na nagra-run 'yung tinatawag na Journalism AI. So they've been doing this for quite some time already. So obviously, 'yung literature ay medyo mas nakatutok sa Global North or 'yung experience sa West.

So if you look at AI and journalism, base sa mga study na ito, maraming gamit siya roon sa buong entire process. Ang tawag nila the news-making process from finding stories, producing stories, and presenting stories.

So kaya rito papasok 'yung mag-i-start tayo, of course, with the transcriptions, and then, pupunta tayo roon sa mga tool na maaaring makapag-process ng data or makapag-sift through ng data. So palagay ko isa ito roon sa among the most, parang powerful na kayang gawin ng AI because maaaring may mga story na hindi natin makikita kung manually natin gagawin. But with the use of AI tools, mas mapapadali 'yung ating paghahanap for stories.

Maaaring rin siyang makatulong sa distribution or presentation of stories. Maaaring ang puwedeng context niya ay, for example, for a lot of reporters, and especially maybe in the Global South where maliliit 'yung mga newsroom natin, oftentimes, ang context 'pag Global South, multitasker 'yung mga journalists.

So maaaring pagod na pagod ka na sa pagbuo ng story mo, pero ikaw pa rin 'yung gagawa ng social media, for example. So isa 'yun doon sa mga finding na 'yun. Maaaring puwede rin makatulong 'yung AI doon sa nasa huling bahagi noong production process, in journalism. So—

HOWIE: But balikan natin 'yung sinabi mo tungkol sa story production, nakatutulong sa pagbuo ng story, sa research, sa reporting. Could you have examples kung paano nakatutulong ang AI? Have you come across any good reporting where nakatulong 'yung AI?

KAROL: Actually, noong 2019, medyo hindi pa ganoon karami 'yung examples of AI-aided investigations, for instance. Pero noong nag-pandemic, malaking impact ito sa atin when 'yun nga, sanay tayo na we go out to field. So ngayon, forced tayo to move, do with what we have na may mga restriction tayo, mobility restriction.

So ang napansin ko roon sa research ay nagkaroon tayo ng maraming examples at isa na rito ay 'yung mga pino-produce ng International Consortium of Investigative Journalists or ICIJ. So gumagamit sila ng, ang version nila 'pag pinag-usapan natin specifically, anong AI tool, 'yung ginamit nila, they use machine learning to detect patterns.

So ito ay if you have a lot of records, 'yun 'yung ginamit nila, computer vision to extract data. So ibig sabihin, if let's say, we're talking about offshore leaks, for instance, na may nakita ka na pangalan ng mga Pilipino roon sa kanilang data set. So 'yung computer vision, for instance, puwede niyang i-crawl 'yung millions of data and documents para mahanap saan lumalabas 'yung pangalan na ito.

So ibig sabihin, mas makikita natin. So ano 'yung sinasabi nu'ng malaking leak na ito about this particular person? So for example, doon sa ICIJ na project na 'yun, PCIJ took part in 'yung Paradise Papers. So ICIJ is the same organization that came up with the Panama Papers, for instance. Pero doon sa Paradise Papers at saka 'yung FinCEN files, isa ito roon sa projects or collaborations na nag-participate ang PCIJ.

So essentially, when we were logging in doon sa repository, we were using AI to crawl 'yung mga document and data. So balikan ko lang, Howie, doon sa Paradise Papers, ang lumabas doon na mga pangalan, for instance, may ilan tayong mga member of government.

HOWIE: So you're describing the method. In other words, AI can help the reporter go through lots and lots of data na normally in a previous era, you would have to look at with a naked eye. It would take you days to look through maybe millions of data points. AI can do it in minutes or even seconds and produce the results that you want.

KAROL: The ICIJ stories have resulted in some of the most impactful global collaboration.

HOWIE: You mentioned the Paradise Papers?

KAROL: So essentially po, ibig sabihin may mga government official na gumagamit pala ng offshore havens.

HOWIE: To park their money.

KAROL: Yes, makapag-evade ng pagbayad ng mga tax or may mga activity sila na idaraan sa isang tax haven para 'yun nga, essentially, hindi makapagbayad ng taxes. Tapos may mga nahanap rin sila roon na apparently 'yung ibang activities na ito ay related pala na may nagsu-supply ng arms. So medyo related siya to war as well. So may mga ganoong finding itong investigation ito na again, 'yun nga, it would have been probably impossible to find without the use of AI.

HOWIE: Okay. Kasi somebody listening might be telling themselves, "Can't you use the find function sa mga Excel sheet or, 'di ba?" I mean, why would you need AI to find the names which are there in the data na puwede mo naman i-search? Ano 'yung advantage ng paggamit ng AI in a project like that?

KAROL: Well, una, we're talking about millions of files. So it would really be impossible to do that na magko-control-F lang tayo, for instance. So obviously, malaking bagay siya na 'yung tedious process na ito ay makababawas talaga ng oras. At pangalawa, 'yung accuracy rin. Kasi kung gagawin natin 'yun manu-mano, it would be really difficult kung hahanapin mo lang siya isa-isa.

And pangatlo rin, maaari rin ibang konteksto siya. So I'll give another example. So this was a story done by Armando Info two years ago. So this is in Southern Venezuela. So nag-start sila with the lead na maraming illegal mines na nag-o-operate in Southern Venezuela. Now, ang context pala roon, these are places na obviously hindi pwedeng mapuntahan. May mga gang doon. So it would be dangerous for a journalist para i-plot ito kung nasaan ba 'yung mga illegal mine.

So ang ginawa noong team, this was led by Joseph Poliszuk at saka 'yung kaniyang partner si Maria Ramirez, pagtingin nila sa satellite imagery, ang isa palang puwedeng identifier ng minahan ay mayroon siyang airstrip. So obviously, kasi 'yun nga, para makuha 'yung supply, kailangan may pagkukunan, kailangan may magla-land doon 'yung eroplano. So 'yun ang ginamit nila.

So they used AI para kumbaga, they fed into the system na 'pag may ganitong airstrip, it's potentially a mine. So 'yun 'yung ginawa nila. So from just a few mines na alam nila nag-e-exist, they were able to track na ganito pala 'yung extent ng illegal mining in Southern Venezuela. So of course, again, nagamit 'yung AI para kumbaga to illustrate the extent of the problem.

It's not just a matter of manu-mano ba siya, pero mayroon kang danger, may risk sa journalist if you would do it, you would go to field. So that's another example.

HOWIE: Okay, nabanggit mo kanina, Karol, si Jaemark. You're referring to Jaemark Tordecilla, who is a former editor-in-chief of GMA News Online. He was also previously with PCIJ, but currently, he's at Harvard as a Nieman Fellow. And one of his projects there has been to develop AI tools for investigative journalism. Have you had a chance to chat with him, talk about this with him, and actually try out his tools?

KAROL: Yeah. I've tried the COA Beat Assistant. And then I think ngayon may bago siyang ginagawa na ang tinitingnan naman ay budget records. I think maganda 'yung ginagawa ni Jaemark ngayon kasi 'yun nga, ina-align niya roon sa paano kaya natin ito magagamit ng mga Filipino journalist, in particular to you know, ano ba 'yung usual na pain points 'pag nagre-report tayo? So obviously, malaking bagay 'pag usapang budget. So obviously, 'yung COA reports—

HOWIE: Commission on Audit.

KAROL: Yeah.

HOWIE: Commission on Audit reports kasi—

KAROL: Commission on Audit reports—

HOWIE: Yeah. They often flag misuse of money or kind of suspicious uses of money, et cetera. So paano ginamit 'yung AI roon sa pag-imbestiga o pagsuri ng mga Commission on Audit reports?

KAROL: So doon sa ginawa niyang tool, ang naging malaking tulong noon ay sa paghahanap ng leads. So puwede kang magtanong na 'yun nga, ano ang key findings from this particular Commission on Audit report? Now, obviously, of course, for journalists, it doesn't mean na kung ano man 'yung ilabas ng tool ay 'yun na 'yung ating ire-report.

Pero palagay ko, ang laking bagay noon na as opposed to reading the entire thing, which ideally, sana ganu'n nga, but alam din naman natin na very technical siya. Ang laking bagay nu'ng may pagsisimulan tayo na lead. So for example, I was looking into Marikina City's Commission on Audit report.

One of my students noong previous semester, they were looking at 'yung use of special education fund. So when we tried the tool, lumabas nga na 'yun, mukhang may issue roon sa special education fund doon sa Marikina City. So doon sa proseso, nakatulong siya sa paghahanap ng story. Tapos 'yung time mo na igugugol sa paghahanap ng story, now, ngayon, puwede mo siyang gamitin para mas i-analyze anong nangyari doon sa paggamit ng special education fund in Marikina City.

HOWIE: Okay, so this COA Assistant, that's what Jaemark has been calling it or that's what you referred to it as. It's in the public domain. I mean, it's free. I mean, kung may nakikinig ngayon na journalism student or anyone who wants to try out the tool, it's just out there. Ise-search lang and makaka-access ang tao.

KAROL: It's just out there. When we were using it for the Marikina COA report, 'yung language nu'ng Assistant on the conclusive side... So, of course, as a journalist, medyo ano tayo roon, dapat mag-iingat because we haven't even seen the record itself at ano 'yung pinagbasihan noong COA Assistant. So siguro, 'yun nga, we know that this is public pero si Jaemark Tordecilla rin naman very mindful din siya na 'yun nga, 'pag gagamitin, of course, we also have to check the original file.

HOWIE: Don't let AI do all of the work, tapos ipapasa na lang natin sa publiko. I mean, kaya assistant lang 'yan. It's going to, it's supposed to just assist us. So whatever conclusions it has, tama ka, we should be skeptical, we should question it, and, you know, we should come up with our own conclusion maybe aligned with the inputs of AI, pero dapat ay galing sa sarili nating utak din 'yun.

KAROL: Kaya 'yung verification ay hindi pa rin dapat mawala. Sa UP, we invited him to speak about 'yung COA Assistant. At naging importanteng punto rin 'yung pinag-usapan namin ng shortcuts. Kasi obviously for us journalists, bago dumating si COA Assistant, which is this AI tool, we've been able to experience 'yung talagang nagbabasa ng COA report.

Kahit hindi man siya productive pero ang laking bagay noon sa atin na maa-apply rin natin sa ibang reporting. Kasi 'yun nga, alam natin na ano ba 'yung composition ng COA report, paano ko ito i-a-analyze, or who else should I speak with para mas makapag-explain nu'ng COA finding.

So 'yun din 'yung pinag-uusapan namin na dapat mindful rin tayo of 'yung mga possible shortcut na me-experience if we use AI tools. So dapat punan natin 'yun by doing our own research, of course, and verification.

HOWIE: Okay. Earlier, you mentioned that people, journalists especially, have been using AI to transcribe interviews. Our profession involves a lot of interviewing and transcribing of interviews para hindi tayo magkamali o hindi tayo nakabase lang sa notes natin or our memory.

Minsan it takes hours and hours to transcribe interviews dati. Ngayon, more and more of us are using AI tools to transcribe. What can you tell us about this? Accurate ba 'yan? 'Yan ang isang malaking tanong.

In terms of the language, kaya ba nila 'yung ito, halimbawa, 'yung pag-uusap natin ngayon, Taglish? At saka alam mo, sa Pilipinas, maraming wika, marami, maraming nagbi-Bisaya, Ilocano, et cetera. So ano rin ba 'yung limits nitong transcribing and ano 'yung potential niya?

KAROL: Well, of course, compared to maybe, you know, several years ago, I think, ang laki nu'ng leap ng improvement ng mga transcription tool ngayon. So du'n sa ilan ko na na-try, useful siya in a sense na kasi hindi ko naman gagamitin lahat 'yung buong transcript sa mga story but at the very least, na-identify ko na, "Ah, ito 'yung may mahalaga siyang sinabi."

Pero I always make it a point to listen back dun sa audio recording kasi may limit rin because as you said, hindi pa siya ganu'n kahasa 'pag Filipino language. Common rin 'yan sa let's say, Bahasa. So parang wala pa ngayon na tool na talagang kuhang-kuha niya 'yung Filipino or even more so 'yung mga dialect.

Pero ang observation ko rin, may naka-capture siya na Filipino. So kumbaga, helpful, helpful pa rin. Pero 'yun, may mga limit pa rin. Hindi rin siya very tedious for, you know, let's say, an intern or a student to do the transcribing. So malaking bagay 'yung tool na 'yan ngayon.

HOWIE: I myself, I transcribe also. So lahat ng journalist, dinaraanan 'yan. So it's made the job of journalism a bit easier. But 'yung flip side nitong co-opting AI, 'yung seeing how it can benefit our profession and other professions, 'yung flip side niyan ay 'yung investigating AI naman. Ano bang kailangan imbestigahan diyan?

I noticed in your presentation at the conference last week, you mentioned an AI Accountability Network. So making AI accountable. What should we be watching out for? What should we be careful about pagdating sa AI?

KAROL: So since last year, I joined as an AI Accountability Network Fellow of the Pulitzer Center. So ito po ay second cohort pa lang. Bago pa lang siya, ano. But basically, itong grupo pong ito ay we're working on AI accountability stories. So essentially, what we're looking at ay 'yung mga ginagamit ba nating tools or services if they use AI or if they use 'yung tinatawag natin na algorithms, tinitingnan natin kung paano ba nito nai-impact 'yung ating daily lives? And whether or not may nakasilip ba sa mga algorithm na ito, ano? Or regulated ba itong mga algorithm na ito?

Kasi kung tinitingnan natin, hindi ba 'yung mga produkto like food products, cosmetics, they go through rigorous testing, di po ba bago mo ilabas sa market. Pero pagdating sa technology, medyo roon po tayo maraming gaps. So kaya maaaring hindi lang tayo familiar but maybe a lot of the tools that we use now on our smartphones, 'yung mga app natin ito, they actually use algorithms na hindi natin alam na who made those decisions, ano?

And 'yun nga, ano ang 'yung impact niya sa atin, base kung paano natin siya ginagamit. So parang ito po 'yung 'pag titingnan natin, ang nakausad dito sa reporting, well, obviously, reporters in the US, in the UK, kasi 'yun nga, they look at Silicon Valley, ano, naging parte ito nu'ng tech reporting, 'yung AI accountability.

Pero malaking, malaking bahagi pa nito ang hindi pa nae-explore, especially in the Global South, and of course, the Philippines, kasi ang laki rin pala ng role natin sa AI. So kanina 'yung nabanggit ko ay investigating algorithms or 'yung tinatawag na black boxes pero kung titingnan rin natin, paano ba gumagana 'yung AI? Ito pala ay dahil we carried out a lot of the digital work na kailangan for AI systems to work.

So kaya rito lalabas 'yung term na data labeling, 'yung mga tinatawag na platform work. So ito 'yung parte rin ito nu'ng AI accountability reporting. So 'yun po 'yung tinitingnan ng mga reporter na part nitong network na ito. Tapos kasunod rin naman nito, ayun nga, ay magkaroon ng training para there would be more AI accountability reporters.

HOWIE: Yeah. Actually, I recall one of the slides in your presentation last week was about digital sweatshops in the Philippines. Itong mga kumbaga business processing center in the Philippines. Based in Cagayan de Oro, itong digital sweatshops na 'to.

According to the Washington Post article, talagang nag-research nang malalim itong mga reporter diyan. And then, ang sinasabi ay there are actually a lot of people all over the world, and in the Philippines, there are two million daw working in these so-called digital sweatshops at very low pay.

They are doing kumbaga, quality assurance or quality control. 'Yung sinasabi mo kanina, data annotation. Essentially, kasi for AI tools to work, they need lots and lots of data, kung saan-saan nanggagaling. But the data also has to be good data. Hindi puwedeng dumugin ng mga kasinungalingan 'yung data na ginagamit ng AI.

Kaya may mga worker sa Pilipinas na nagle-label nito. And reviewing it for accuracy, maybe deleting some or recommending some for deletion or non-use by AI. So hindi alam ng marami 'yun na may ganu'n pala.

So isa 'yun sa mga shinare mo roon sa presentation mo na that there's a very human element behind all of these AI machines. Itong machine learning ng AI, ang daming human inputs diyan.

KAROL: It's interesting because when we talk about AI, as you mentioned, ang laki nu'ng human component pala noon. For example, 'yung mga nakausap ko, as I understand, these are for self-driving cars so they mark road signs. Parang ano naman 'yun, manonood sila ng video na may daan siya, I think it's a Japanese road, 'yung isa kong nakita. So you can think, it's probably for Japanese 'yung client naman noon.

So magma-mark ka ng sign na may pedestrian dito or bridge. So ganu'ng kind of work. That's what you do the whole day. May isang example naman, nakuwento ito sa akin nu'ng kasama ko ru'n sa fellowship sa Brazil naman na ang tini-train na AI ay para doon sa 'yung robot na vacuum. Pero ang pinapagawa ay you have to take photos of dumi ng aso. Kasi apparently, siyempre 'yung AI, hindi naman niya alam na dumi pala 'yun. So maaaring linisin niya.

So kailangan mo mag-input, mag-feed ng data na ito 'yung hitsura ng dumi. So dapat hindi mo siya lilinisin. So may ganu'ng type of data labeling works. So ito 'yung types of work na apparently 'yun nga ano, a lot of countries in the Global South, Philippines, Latin America, India also, ay ito 'yung ginagawa na trabaho.

So malaki 'yung role natin doon sa supply chain. Tapos tiningnan natin 'yung labor concerns which was brought up in the Washington Post report. Siyempre maliliit lang 'yung bayad niyan. Tapos papasok naman tayo roon sa usapin ng platform work. Kasi 'yun, hindi naman sila regular employees. So nagiging subject rin to exploitation 'yung mga digital worker.

HOWIE: Sabi nga roon sa article, hindi lang mababa ang suweldo, madalas delayed ang sahod nila. At may mga instance na hindi na binabayaran. But dahil nga kulang sa trabaho sa maraming lugar sa Pilipinas, talagang tinitiis na rin ng maraming Pilipino.

But, you know, for those listening and considering work like that, maybe that's something to think about. I want to ask you, kasi nabanggit mo kanina, one of the questions about AI, is it going to cause the displacement of workers? Inevitable 'yan sa mga bagong technology, right?

Pero lagi may pangako na it's going to create jobs so it's going to displace workers but they can be retrained for other kinds of jobs that are created by new technologies. To what extent is that true with AI?

KAROL: So my research on this is still ongoing but so far i'm looking at obviously impact on the BPO sector. May mga lumalabas ng reports na sabi nila 1.8 million jobs raw 'yung mawawala with the adoption of AI. But so far medyo complicated 'yung picture because you will hear, you know, stories na they use AI pero that would mean they would do other jobs so or maaaring may maapektuhan but not the full-time or the regular employees.

Even more so magandang tingnan because obviously, you know, malaking bagay 'yung outsourced work sa Pilipinas. So kung isa ito roon sa maapektuhan ng AI, I think magiging malaki rin 'yung impact noon sa Pilipinas. But so far I can't say anything conclusive yet kung 'yun nga ano like let's say 'yung mga pine-predict na 1.8 million ay kailan ba 'yan or have we experienced it? But so far 'yun, ang consistent ay may ibang job na ginagawa kasi 'yun nga maaaring AI na 'yung gumagawa ngayon.

HOWIE: Well, Karol, nabanggit mo nga 'yung transcribing no so a lot of transcribing around the world is already being done by AI tools. This particular interview will also be transcribed with an AI tool but we have not let go of our transcriber. Ang trabaho niya ngayon ay taga-review ng transcript kasi hindi naman 100% accurate no 'yung 'yung transcribing tool na ginagamit namin.

So kailangan pa rin ng tao na magre-review. As far as I know wala pang nadi-displace despite the fact na naging mas efficient ngayon 'yung transcribing but it's it's still far from perfect. So kailangan pa rin ng tao.

So at least, sa larangan namin 'yung paggamit ng AI ay wala pa rin nawawalan ng trabaho. But currently it's just helped in our production.

I wanna pivot a little bit to something that's being discussed a lot, 'yung deep fakes. Deep fakes. Itong mga video ng totoong tao na nagsasalita ng mga kung anu-anong nonsense na minsan delikado 'yung mga sinasabi nila. May panganib na hindi talaga nila sinabi. So that's why it's a deep fake. It's convincing.

Mini-mimic 'yung boses nila. Minsan sumusunod pa 'yung bibig, 'yung lips. Nakakaloko. So share your thoughts about this. What's the potential of that to cause great harm?

KAROL: I think, sa umpisa, Howie, 'pag tiningnan mo kasi, hindi ba parang halatang-halata? At first glance, parang you wouldn't take it seriously. Kasi kahit gaano pa ka-smooth siya, alam mo pa rin na parang may something off. Pero that's me, 'no? From the point of view of maybe familiar sa ganito.

I think it's a legitimate na parang source of concern 'yung mga ganitong lumalabas na material, lalo na na may experience tayo with several elections already. I think, our first one would be Argentina and then India also na kung saan ginamit 'yung mga tool na ito para makapag-produce ng, you know, you have videos of a politician or a candidate running na puwedeng ipadala sa mga voter o kaya may mga robo-call na magsasabi sa 'yo na "Huwag bumoto," 'no?

So 'yung mga ganitong type ng content, dapat tutukan natin in a sense na saan ito nanggagaling. Kasi palagay ko, malaki 'yung puwede niyang maging impact, especially 'yun nga, 'pag pinag-usapan natin in the context of an elections.

Tapos siyempre, of course, dadagdag pa 'yan with the whole disinformation problem natin. Hindi sa sinisisi natin 'yung tao, pero kasi kung 'yun 'yung ipinakita, maaaring maniwala talaga tayo roon sa mga ganitong material.

So yeah, I think it's a cause for concern. Lalo na 'yun, again, in the context of elections, for instance.

HOWIE: Okay. In the context naman of the university where you teach. Alam ko, malaking issue rin 'yan sa academe kasi ang laki ng temptation para sa mga estudyante. Even siguro professors also have to write and research. So ang laki ng temptation na gumamit ng AI.

At ano ba 'yung boundaries diyan? I mean, you don't, I'm sure you do not tell your students not to use AI. Kasi, tinuturuan mo nga ng mga tool, 'di ba? Pero hanggang saan ba dapat sila gumagamit ng AI?

KAROL: Right now, ayun, pinag-usapan 'yan sa academe, ano ba ang policy on AI? As I understand, mayroong nilabas ang university pero parang hindi pa siya as specific to, you know, like, sa production ng assignments, for instance.

Sa akin naman, personally, worried ako doon sa pag-shortcut. Hindi ko sinasabi na huwag natin i-explore ang AI tools. Kasi kakasabi ko nga lang kanina, co-opting AI. Pero doon sa usapin na nasa university tayo at nag-aaral tayo, especially in journalism, parang, ang laking bagay nu'ng ikaw 'yung nagbasa talaga, ano, to synthesize ano 'yung findings and everything.

Kaya 'yun, this is something na hindi ko ine-encourage 'pag magsusulat ka, of course, of a report. Especially, when we talk about original reporting kasi obviously, if you use AI, saan niya 'yun ipu-pull, ano? From material na already published. So you still have to do your own interviews, make sense of the information that you got.

HOWIE: Okay, Karol, as a professor, can you tell if your students' submissions were the product of AI? May paraan na ba ang mga teacher, mga professor ngayon para malaman kung hindi 'yung original work, pinagawa niya sa chatGPT or another AI tool?

KAROL: I think, it's a little bit hard to detect. I know may mga tool na makakapag-detect. Kaya lang, as I understand, 'yung chatGPT, nag-i-improve rin siya, eh. If you, let's say, mag-enter ka ngayon ng one query, and then you enter it again later, it won't give you the same results. So technically, if we follow that, if I enter the same, hindi ko mata-track mismo 'yung nahanap, 'yung sinubmit nu'ng estudyante.

So dito, titingnan mo 'yung portfolio niya, ano, 'yung work portfolio niya para ma-check mo kung, "Wait, ba't parang biglang nag-iba 'yung output niya ngayon when alam ko na ito 'yung quality of work na sina-submit niya. So 'yun, medyo mahirap siyang i-detect.

HOWIE: So kailangan mo ng reliable benchmark. Kunwari, sa umpisa ng semester, ganito na 'yung output, at alam mo na 'yun ang trabaho niya kasi sa, you know, sa classroom niya ginawa. Nakita mo. And then, later on, may assignment, ang sinubmit, napakalaking improvement or ibang-iba 'yung estilo or iba 'yung vocabulary words na ginagamit. Ganu'n ba? You really have to just trust your judgment on it, 'no?

KAROL: Saka sa discussion rin, kasi syempre, what I do is I ask them to report about their outputs also. So siyempre, dapat 'pag tinanong mo, pina-elaborate mo, dapat madi-discuss din nila 'yun fully. Kaya lang, 'yun nga, kasi ang context ko rin naman, dahil 'yun nga, investigative reporting, dahil may impetus on getting original information, medyo mahihirapan sila roon na gumamit ng AI kung may ipapa-summarize sila or may iaasa sila sa AI tool na gawin 'yung parte nu'ng report na 'yun. They would have to interview people for that, or analyze data for that.

HOWIE: Okay, my final question, Karol, are you more optimistic than pessimistic about AI? Kasi maraming mga scenario, maraming opinions for the future, may mga nagsasabi na, you know, even technology leaders may nag-i-issue ng warning na, "If we're not careful, AI will take over the world," you know, it's going to take control of institutions, magkakaroon ng super intelligence in the near future, where they're going to disregard human judgment and decision making and make decisions on their own.

On the other hand, may mga nagsasabi, "No, 'yung AI ang magse-save sa humanity." It's going to enable us to solve problems, et cetera. Where do you stand on this?

KAROL: Usually ganyan 'yung conversation, either doomsday or techno-solutionism nga na this is the answer. Pero I think, if titingnan rin natin 'yung history pagdating nu'ng print and then internet, parang the reality is somewhere in between. So sa akin, siguro, I am cautiously optimistic because I've been using the tools and na-experience ko 'yung advantages niya.

So 'yun din sana 'yung ma-practice rin nu'ng iba para we're able to harness AI to improve our reporting. Kaya 'yun nga, co-opting AI. However, 'pag may mga usapin tayong alignment, 'yun, 'yung medyo nakakakaba, ano 'yan, alignment na parang, ito na 'yung usapin nu'ng general AI, na may mga depiction tayo na capable siya of reasoning and all.

Imo-monitor din natin ito kasi a lot of these things are happening na behind the scenes. 'Yun nga, 'yung paggamit ng mga algorithm. So siguro ang kaakibat naman noon ay dapat ay we also investigate and we also seek accountability dito sa mga AI system na ito.

So in that regard, kung gagamitin man 'yan for purposes to sway elections, for instance, kailangan mag-raise ng awareness in that regard. Kasi definitely, that's not how we want AI. We don't want AI na maging ganu'n 'yung impact sa atin.

HOWIE: Okay. So you are cautiously optimistic and the keyword there, of course, is cautious. Good advice. So we want to thank you on that note, Karol. Thank you for sharing and for your valuable research on this transformative technology. We're looking forward to your book, Prof. Karol Ilagan, journalist, researcher, at faculty member sa UP Diliman. Maraming salamat at mabuhay ka.

KAROL: Thank you, Sir Howie. Maraming salamat po.