TAKEAWAYS
If 2022 belonged to TikTok, then 2023 is the year of ChatGPT. The current hype around the tool – and generative artificial intelligence (AI), the tool that powers it – might make it seem like it is a new phenomenon but this is far from the truth. It is merely building on technology that has been gradually creeping into our lives for years. Think about your email trying to suggest what you want to type next, or Spotify building a playlist that you may like. And if you pop into your mobile phone’s Photos folder, you’ll find that it has likely identified your friends’ faces and grouped their photos together for you.
What has changed though, is the enormous progress in computing power and data analysis. The popularity of ChatGPT has brought these developments to the fore. The tool, which was launched in 2020, can generate human-like responses to a conversational prompt or message. Earlier this year, it leapfrogged other programmes like Instagram and TikTok to become the fastest-growing web platform in terms of the number of active users.
It’s not the only AI tool around. Others include Midjourney, which can be used to create realistic images based on a prompt, and DeepArt.io, which allows users to transform their photos into artistic styles inspired by famous artists. Although their outputs may be different, these tools work the same way. Grasping how AI works is crucial as it helps dispel certain persistent myths – including one that such tools have a mind of their own and can take over the world. Such scenarios can easily be seen as the stuff of science fiction once users understand that AI tools are probabilistic engines that come up with predictions on what a user is asking, based on huge databases of past data. These databases “train” these tools to learn patterns and correlations, which are then used to create content that is both realistic and lifelike.
ChatGPT’s impact is being felt beyond the virtual: it’s changing the way we work, heralding a new era of creativity and productivity. According to a survey cited in Business Insider, 43% of professionals have used tools like ChatGPT at work. Even Singapore’s public service has embraced some form of AI, which officers can use to help them in research and writing. However, that has been accompanied by a directive to ensure the work they eventually submit is accurate and in line with copyright laws.
Accountants, too, have tapped on ChatGPT to work smarter and more efficiently. Such tools eliminate the need to scour through research to learn about tax regulations and accounting standards. They can also assist accountants to calculate ratios, perform simple financial forecasts and analyse data patterns.
But even firms that have adopted such tools should err on the side of caution as ChatGPT is not a substitute for accounting software or advanced financial models. Instead, it should be viewed as an assistant, to be used alongside professional judgement and training. There are also ethical and privacy issues to consider. “One thing to be wary of is feeding privileged client information into (ChatGPT),” said Ben Bilsland, partner and technology sector expert at RSM, a leading British audit, tax and consulting firm. “That’s a problem for accounting firms,” he added, speaking to Accountancy Age.
OpenAI, the Microsoft-backed tech company behind ChatGPT, has sought to reassure professionals with new features, to allay concerns about data security and integrity. In March, it introduced a feature that allows users to indicate if certain conversations should not be used to train future models. However, they will still be stored in the company’s servers for 30 days. This may not go far enough, although OpenAI does have plans to roll out a business version of ChatGPT, which will offer better data security and privacy. It remains to be seen how these will comply with the strict standards on data privacy and security that the accounting profession must adhere to.
Other less obvious use of ChatGPT in finance and accounting includes marketing oneself and one’s services. Those looking to grow their social media footprint can turn to ChatGPT to generate snazzy posts about trending topics. People who are less confident of their writing can also use ChatGPT to vet their emails and improve their tone and clarity.
While some professions have embraced ChatGPT and other AI tools for their efficiency, other fields view them with suspicion, concerned about how it might take away, rather than help, jobs. Some have already sounded alarm bells. In March, Goldman Sachs economists estimated that as many as 300 million full-time jobs could be automated in some way because of advances in AI. But at the same time, World Economic Forum research suggests AI is projected to create around 97 million new jobs, potentially countering workforce displacement concerns.
However, some jobs will likely still be affected, especially among professionals in creative fields such as writers, designers, photographers and musicians. AI-generated content – which ChatGPT creates in seconds – may become more prevalent, leading to a reduced demand for human creators. Professionals in customer service may also be affected as already, ChatGPT-powered chatbots and virtual assistants are being used to handle customer inquiries and support requests. As the language models they are trained on improve, they may be able to handle more complex customer interactions, potentially reducing the need for human customer service representatives. Another strength of ChatGPT is its ability to quickly process and analyse vast amounts of data, which may impact jobs in fields like data entry and basic data analysis. Many administrative tasks, such as scheduling, document processing, and data entry, can also be automated using similar tools. This could impact administrative assistants and similar roles.
This is not to say that AI tools like ChatGPT do not have their drawbacks. One of these has to do with accuracy. There have been numerous instances of ChatGPT presenting inaccurate data, for example, when it “hallucinates” and generates output that sounds plausible but is either factually inaccurate or contextually irrelevant; in some cases, the output can even be dangerous. Part of its popularity stemmed from its promise of content guardrails that promised to limit hate speech, violence, misinformation and instructions on how to do things that are against the law. However, the perception of that illusion was swiftly shattered as users discovered methods to circumvent content moderation. As reported in The Guardian, instances arose where ChatGPT could be manipulated to produce derogatory comments about women and express sympathy towards war criminals like Adolf Hitler. Generative AI can also be easily used to generate malicious code that can launch cyberattacks or fake news campaigns, and impersonate others by generating lifelike images.
Fortunately, not all is lost. Governments, such as in the European Union, have increased their scrutiny of ChatGPT and are mulling legislation to regulate its development and use. Closer to home, the Infocomm Media Development Authority (IMDA) has spearheaded the AI Verify Foundation (AVF), a group comprising at least 60 global industry players. Including tech giants like Microsoft, Google and Meta, AVF will tackle pressing issues in AI such as bias, copyright and its susceptibility to lying, said Minister for Communications and Information Josephine Teo, at the Asia Tech x Singapore Conference in June.
Outlining Singapore’s approach to AI, she said that the government would work with industry to steer AI towards beneficial uses and away from harmful ones. “This is core to how Singapore thinks about AI,” she told an audience of several hundred tech professionals. “In doing so, we hope to make Singapore an outstanding place for talent, ideas and experimentation. A strong desire for AI safety need not mean pulling the drawbridge to innovation and adoption.”
A YEAR TO REMEMBER
A whizz through ChatGPT’s eventful year so far:
- January 2023: ChatGPT reaches 100 million monthly active users – a record-breaking feat.
- April 2023: Italy becomes the first major jurisdiction to block ChatGPT, citing data privacy concerns. It resumed service in the country later that month.
- May 2023: ChatGPT goes mobile, with an iOS app. An Android version opens for pre-orders two months later.
- May 2023: Minister for Communications and Information Josephine Teo confirms to Parliament that the Public Service has introduced guidelines for public officers using technologies like ChatGPT to draft documents.
© 2024 Institute of Singapore Chartered Accountants