Filipinos and others living in the GMT+8 timezone woke up on Wednesday, March 15 to Facebook posts and Twitter threads about the announcement of GPT-4, OpenAI’s latest multimodal generative artificial intelligence (AI) model. Much will be said over the next few days in the media about how this could change our world, but by the end of this piece I hope to remind everyone about one way the new technology could end up impacting the Philippines in particular.
For those who have not been following the latest developments in artificial intelligence, generative AI is AI that can create “new” text, images, and other content in response to its understanding of prompts. Some months ago, OpenAI’s earlier product ChatGPT – a service that uses generative AI for text – promised to write responses to people’s questions written in natural English. It demonstrated unprecedented capability in writing essays, answering complex questions, and remembering things that users told it earlier in a conversation. It took the world by firestorm, sparking conversations about the future of writing, productivity, and education. Teachers found students using ChatGPT to write essays not only in high schools across the world but even at the University of the Philippines. Some have suggested that its ability to draw from a large pool of knowledge would threaten occupations like law.
This shock and awe, however, was tempered by those who pointed out ChatGPT’s limitations. Firstly, it can only understand text. Second, because it was trained on Internet data from 2021 and earlier, it cannot fetch updated information. Lastly, it sometimes “hallucinates,” or creates answers that are wrong but which seem to make sense.
Not even a year has passed since ChatGPT was unleashed into the world, and GPT-4 now promises significantly improved performance and problem-solving capabilities. One important thing about GPT-4 is that it is multimodal, which means that it does not only understand and generate text but can also write based on its understanding of other kinds of data such as images. The GPT-4 technical report shows that it can solve physics questions that rely on diagrams written in French; explain jokes in memes; and produce impressive scores on many standardized tests.
We learned that the new AI feature in Bing – a search engine by Microsoft, which helped fund OpenAI – revealed weeks ago was using GPT-4 the whole time. The use of GPT-4 in Bing is especially interesting, because as Bing is connected to the internet, it addresses the second limitation of ChatGPT and can fetch updated information. As a result, Bing can answer questions about news and current events much better than ChatGPT can (if not always correctly), and it can cite its sources, too.
GPT-4 still has its limitations. The GPT-4 in Bing can still give incorrect answers and, on certain occasions, does not look like it has emotional intelligence. But the improvement between the original version of ChatGPT and GPT-4 has been significant, and I think it’s time that we stop downplaying its potential to impact our world.
I agree with many who believe that the unleashing of generative AI has been done in an ethically questionable way. Artists in the Philippines have been very vocal on Twitter about how DALL-E, a generative AI that produces images based on existing art, might threaten the livelihoods of artists. As earlier mentioned, teachers are afraid that this would hurt students in the long run as they rely on AI to complete their assignments rather than learn to think for themselves. Their perspectives are just a few among many who think that the emergence of generative AI has been too sudden and too irresponsible, and it explains why many are upset at a tech industry that seems to “move fast and break things” before considering how their actions impact society. I agree with them.
But Pandora’s Box is open, and it’s important to have conversations about how this new form of AI can fit into our society, and how society can grow around AI, instead of leaving those decisions up to people who stand to benefit, in terms of money and power, from the adoption of AI.
The most obvious way generative AI will create value for businesses is to reduce the need for workers to answer questions and process information. Upon the release of GPT-4, many tech observers – especially from India, our major competitor in the business process outsourcing (BPO) space – are predicting that the technology will soon be deployed in customer service settings.
One of the examples in the GPT-4 technical paper is a user asking the model to purchase a compound with similar properties to an existing drug. GPT-4 was shown to be able to not only identify such a compound but also identify a supplier that sells it, and it purchased the item by itself – all based on a single request from a user, without human intervention.
As generative AI can now seek information from many different data sources, including the live internet, and answer questions more helpfully than ever before through not just words but actions, it is not unthinkable that companies will seek to deploy this technology to deal with hundreds of thousands of customers who need help at the same time.
This has ramifications for our business process outsourcing (BPO) sector, which employs 1.44 million workers in a nation of 109 million not only in Manila but also in secondary cities like Cebu, Bacolod, and Iloilo. The sector produces almost $30 billion in revenue and is one of the few industries that expanded in the Philippines during the COVID-19 pandemic. This also has implications for the rising “virtual assistant” (VA) trend that is allowing many across the country to seek work directly from overseas employers.
The concern that AI could threaten our BPO industry is not a new one. In the past couple of years, several companies have attempted to provide AI call center services. While they are not yet impacting employment in an obvious manner, some Philippine-based BPO companies were starting to use the technology. That was enough for some lawmakers to take notice. Last January, Senator Risa Hontiveros asked the DTI to take measures to help BPO workers keep up with advances in AI, saying that if the industry disappears, it would deeply hurt the Filipino workforce as “a pathway to entry into the middle class will also disappear.”
I think the release of GPT-4 should breathe new urgency into this issue. Overshadowed by much of the social media furor over GPT-4’s announcement was the unveiling of one company’s product that already uses GPT-4 in a customer service chatbot designed to replicate many of the tasks done by customer service agents. It is not hard to imagine that as the technology matures at this exponential rate, more and more companies will choose to use customized AI services – especially given that they are now able to provide much of the personalized touch that used to be considered one of the strengths of Filipino outsourcing workers.
While customer service outsourcing is the type that is most obviously at risk to automation, it is not unreasonable to imagine that many of the other knowledge industries in our country, outsourcing or not – such as accounting, web development, and medical transcription – might be at risk, too. Another use case of GPT-4 going viral is its ability to generate usable website code from a sketch on a napkin. While developers will still be needed to turn this prototype into a fully-functioning website, why would a team in the United States or Singapore equipped with AI feel the need to outsource the basic work to another country?
It is possible I am overstating the risk at hand, and I hope that is the case. But I argue that this is a risk we cannot afford to underestimate. People were complacent about AI’s threat to knowledge workers because of ChatGPT’s flaws and GPT-4 shows that progress is being made to patch over them – if hastily and arguably irresponsibly. A common adage in the Philippines is to “hope for the best and prepare for the worst,” and we must prepare for a scenario where it becomes harder to attract foreign direct investment (FDI) into the Philippines.
In 2012, the Asian Development Bank warned that the Philippines is still suffering from a weak industrial base. It was overly reliant on the BPO sector for strong economic growth the country enjoyed during the 2000s. While BPO has given many Filipinos with English skills opportunities to provide for their families without leaving the country as OFWs, regardless of their college degrees or lack thereof, it is also still not accessible to all Filipinos. If automation can reduce the number of workers that companies need to effectively provide these services, then even this source of jobs is at risk. Its domestic economy is still chronically underdeveloped.
So much of our country and its economic development model relies on the BPO sector. It is because of the BPO sector that real estate developers have been able to create IT Parks and their “modern” skylines – the ones that people like to caption on social media as being “New York” in the Philippines. It is because of the BPO sector that cities like Bacolod were able to find new economic futures after the decline of local industries like sugar. It is because of the BPO sector, and of other opportunities like virtual assistant jobs, that people can earn “in dollars” without saying goodbye to their children. Our BPO sector deserves more support than ever, but our country also needs to have more ways of creating opportunities for Filipinos.
What would happen to our knowledge workers in the worst case scenario? Will the BPO sector be reduced to supporting AI, as was the case with OpenAI paying Kenyan workers less than $2 per hour to help improve ChatGPT?
The Philippines needs to act quickly and decisively to mitigate the possible risks around AI. I cannot say I have all the answers – I am merely a student – but in my humble assessment, some of the steps our country needs to take include:
- Economic diversification that aims to improve productivity across multiple sectors, including and especially modernizing agriculture, so that Filipinos can earn livable wages and grow as workers and people instead of simply being exploited by foreign corporations for the lowest possible cost;
- Paying and training teachers and investing in education that goes beyond memorizing facts and formulas to prepare Filipinos for uncertain futures, so that they can find careers that allow them to contribute meaningfully to their communities instead of competing with AI in a race to the bottom;
- More sustainable forms of real estate and economic development – for example, building flexible workspaces, affordable and dignified housing, and facilities for good public transportation – instead of large megaprojects that rely on a certain kind of industry staying the same forever, which is too much to ask of a rapidly-changing world;
- Lastly and crucially, a government that will not chase short-term development goals at the possible expense of Filipinos’ long-term wellbeing and economic futures.
The growth of AI should be scrutinized and if possible regulated, but as it happens it will also force countries around the world to be more conscious about how they develop their labor forces and protect their citizens in an unstable world. The Philippines is no exception. Hopefully, in the coming years we can say that we took our current moment of flux as an opportunity to rethink our social contract and finally develop strong economic fundamentals instead of relying on band-aids. – Rappler.com
Joshua Vargas is a final-year student at Yale-NUS College, Singapore. As a Donor Study Award recipient, the 22-year-old Zamboangueño majors in Urban Studies and minors in Computer Science. He is currently conducting his thesis research on the future of Philippine cities as remote work transforms the BPO industry.
There are no comments yet. Add your comment to start the conversation.