ChatGPT chatbot has taken the world by storm since its launch in November. The artificial intelligence (AI)-powered tool can answer questions in beautifully written sentences, assist programmers in writing code, compose songs and poems, as well as write articles.
Created by OpenAI, an AI lab in San Francisco, ChatGPT made headlines as it surpassed 1 million users within a week after its launch. Software giant Microsoft recently said it plans to invest US$10 billion in OpenAI.
While this generative AI can make life easier for many people, such as allowing teachers to prepare courses and exam papers, the technology could pose a challenge for some jobs, such as content creators and researchers.
The chatbot sometimes gives the wrong information, hampers thorough studying by students, fuels confusion about copyright ownership, and could become a tool for cybercriminals.
Analysts suggest people pursue upskilling and reskilling to use generative AI effectively in the future, utilising the technology to work quicker.
ChatGPT was built on GPT-3 (Generative Pretrained Transformer 3), a language processing AI model developed by OpenAI. It is the largest and most powerful language model ever created, with 175 billion parameters and the ability to process billions of words in one second.
GPT-3 works by pre-training a deep neural network on a massive data set of text and then fine-tuning it on specific tasks, such as answering questions or generating text.
New research conducted by a professor at University of Pennsylvania's Wharton School suggested ChatGPT could pass the final exam for the school's MBA programme.
PCMag.com reported OpenAI is developing a paid version of ChatGPT to support faster response and a broader range of answers than the free version.
Thanachart Numnonda, executive director of the IMC Institute, a research and technology training provider, said ChatGPT is a sequence model of coordinated messages that can create comprehensible and effective messages using high-performance learning techniques as well as synchronising messages between users and models.
It has an AI program using deep learning techniques and it will become better with more usage, he said.
Mr Thanachart said the platform still has some drawbacks, such as limited knowledge of data more recent than from 2021 and usage of a large amount of computing power. In addition, sometimes it can generate false answers.
ChatGPT is still in its early stages, as learning to accumulate more knowledge from users will help it to improve efficiency, he said.
"It may be too early to determine the impact of ChatGPT for human tasks, but what is certain is in 2-3 years, content writers, travel agents, instructors, researchers, and English-language call centres will all see early challenges from AI," said Mr Thanachart.
However, he believes critical thinking and creativity still cannot be replicated by AI.
"Individuals need to build new skill sets and use those tools to develop more opportunities and speed up work," said Mr Thanachart.
Panutat "Jimmy" Tejasen, a tech entrepreneur and software developer, agreed that ChatGPT will make life much easier for people with an easy-to-use, "conversational" user interface. ChatGPT can assist programmers in providing source code that works well with task requirements.
"If you face a programming problem, this AI tech can find you the answers and provide an example of code, which supports task optimisation," said Mr Panutat.
There are still some errors in the platform, but in general ChatGPT can be helpful and increase the productivity of programmers, he said.
"It still cannot replace humans at this time," said Mr Panutat.
He said ChatGPT was built from five years of research. Thailand does not have much research in this field, said Mr Panutat.
It is not easy to make an AI tool that supports Thai language as it requires massive investment, he said. Some potential fields may be chosen to be supported by Thai language in AI, said Mr Panutat.
He said people with good command of English can leverage the powerful tools available from AI technology, which could create a gap between those who lack English skills and cannot utilise the technology and those who can use AI.
Countries with good command of English would gain an advantage from AI, said Mr Panutat, creating a challenge for Thailand where most people only use Thai language.
He said a large data set in Thai language should be used to train machines with powerful computing through a collaboration between universities, large businesses and science parks. Those who can capitalise on AI would be able to access more knowledge and deliver work faster, said Mr Panutat.
"Generative AI will play an important role in changing the world," said Mr Panutat.
Virach Sornlertlamvanich, a researcher on national language processing and social media understanding, especially Thai language processing, said users who embrace AI tools will have an advantage over others. Developers should leverage AI technology to create new innovation, he said.
"Thailand needs to build an environment and infrastructure that can attract tech talents to work together with funding support in order to conduct research that can deliver a meaningful impact," Mr Virach said. "Research like ChatGPT requires a lot of funding and years of development. This requires significant project planning and a pool of top experts that can deliver important technology to drive the country's economy."
Siraprapa Chavanayarn, a lecturer at Chulalongkorn University's Department of Philosophy, said as ChatGPT plans to offer a paid version with more powerful functions, this means those who can pay will gain a greater advantage over others who cannot.
"This creates a kind of digital divide," she said in a recent online discussion on ChatGPT in terms of opportunity and learning challenges, organised by the university's Humanities Information Center.
While Google and Wikipedia provide a wide variety of sources, ChatGPT creates summarised content that needs to be rechecked for accuracy and bias.
"This new tool can help us access knowledge, but it should only be used to complement our work," said Ms Siraprapa.
She said the platform can affect how people acquire knowledge. How well ChatGPT provides answers may depend on how well the questions are created by learners, said Ms Siraprapa.
Students should be encouraged to ask questions in a way that allows ChatGPT to deliver more analytic answers, which can be useful for learners, she said.
NEW LEARNING PARADIGM
Wirote Aroonmanakun, director of Chulalongkorn University's Sirindhorn Thai Language Institute, said at the same seminar teachers can use AI to prepare course papers and design exercises for students.
Teachers may no longer need to gauge students' performance from their homework, which can be done by AI. This means it will become harder to prove whether students did the homework themselves, he said.
"Teachers need to gauge a student's performance based on their understanding. Questions students put to an AI tool can demonstrate whether they understand the subject. AI can glean the answer for students," said Mr Wirote.
"AI will accelerate learning pace," he said. "People need to be smarter to use the tool."
Attapol Thamrongrattanarit, a lecturer at Chulalongkorn University's Department of Linguistics, said while AI can support some parts of teaching, it cannot replace all of them as particular subjects require interactive discussion and critical thinking.
"AI can be a powerful tool to support teachers and learners in the future," he said.
Songphan Choemprayong, a lecturer at Chulalongkorn University's Department of Library Science, said the use of AI to create work can pose a challenge for intellectual property rights.
"As AI extracts knowledge from multiple sources, how can we define 'original content' for a copyright?" he asked.
ChatGPT also added some spice to the cyberthreat landscape as it quickly became apparent that code generation can help less-skilled threat actors effortlessly launch cyber-attacks.
According to Check Point Research (CPR), a cyberthreat intelligence provider, ChatGPT successfully conducted a full infection flow, from creating a convincing spear-phishing email to running a reverse shell, capable of accepting commands in English.
CPR said its analysis of several major underground hacking communities shows there are already instances of cybercriminals using OpenAI to develop malicious tools.
"As we suspected, some of the cases clearly showed that many cybercriminals using OpenAI have no development skills at all," said CPR. "It's only a matter of time until more sophisticated threat actors enhance the way they use AI-based tools for illegal purposes."