ThengHui — Notes

AI Transformation Turns the Impossible Possible

Sat, 14 Jun 2025 08:52:06 GMT

You know, I still remember, I wrote about what AI could not do, 5 years ago... Five years ago, our imagination of artificial intelligence likely stayed within the fictional realms of sci-fi movies. Back then, AI was still seen as a cold tool, proficient in data analysis and executing repetitive tasks. However, when it came to creative ideation, emotional understanding, or complex strategic judgment, we generally considered these areas beyond AI's reach. People widely believed AI couldn't generate natural, fluent text, create art, let alone engage in deep conversations with humans. Yet, in just five short years, the pace of AI development has been astonishing. Tasks once deemed "impossible" are now everyday capabilities for AI. Large language models have emerged in droves, not only writing articles and poetry but also coding and designing products. Image generation AI has turned imagination into reality, creating exquisite pictures from simple text descriptions. Our interactions with AI have also shifted from command-based to conversational; AI has become more empathetic, better at understanding context, and able to provide personalized services. In a mere five years, AI has gone from "cannot" to "can," evolving from a tool into a collaborator, even surpassing human performance in some aspects. This has not only changed our way of life but also redefined our understanding of intelligence. What new surprises will AI bring in the next five years? We eagerly await to see. --- # 五年之變：AI 翻轉不可能五年前，我們對人工智慧的想像，或許還停留在科幻電影的虛構情節。那時，AI 仍被視為冰冷的工具，擅長數據分析、執行重複性任務，但對於創意發想、情感理解乃至於複雜的策略判斷，我們普遍認為是 AI 難以觸及的領域。人們普遍認為 AI 無法生成自然流暢的文本、創作出藝術作品，更遑論與人類進行深度對話。然而，短短五年，AI 發展的速度令人咋舌。當年被視為「不可能」的任務，如今已成為 AI 的日常能力。大型語言模型如雨後春筍般湧現，它們不僅能撰寫文章、詩歌，甚至能編寫程式碼、設計產品。圖像生成 AI 則將想像力化為現實，透過文字描述即可創造出精美絕倫的圖片。我們與 AI 的互動也從指令式轉變為對話式，AI 變得更善解人意，更能理解語境，提供個性化服務。五年的光景，AI 從「不能」到「能」，從工具演變為協作者，甚至在某些方面超越了人類的表現。這不僅改變了我們的生活方式，也顛覆了我們對智能的定義。未來五年，AI 又將帶來什麼樣的驚喜？值得我們拭目以待。

Towards the Future: From Learning Machine Language to Machines Learning Human Language

Sat, 01 Jul 2023 08:52:06 GMT

![AI-generated photo - Posting up my article on a bulletin board in classroom ](img/aiphoto-student.jpg) I still remember my fifth year in secondary school, when our country was promoting the era of the Multimedia Super Corridor. Back then, the internet was referred to as the "information superhighway". I wrote an article and pinned it on the classroom bulletin board, eager to share my thoughts on technology with my classmates. Now, those feelings are resurfacing, and I find myself passionately wanting to write again, haha. As someone who stepped into the tech industry after graduating from university, I still remember the times when we were all learning machine languages. We diligently learned programming languages like C and C++ to communicate with machines and issue our commands. However, the world changes so quick. Now, our goal is no longer to learn the language of machines, but to have machines learn our language, mimicking our patterns of thought. In other words, we are no longer merely commanding machines to carry out our directives, but training them to understand our intentions and even predict our needs. Several years ago, we still have the view that data precedes artificial intelligence but now seems outdated. The emergence of Generative AI has opened new vistas; We can now generate new data directly using AI. This innovative technology offers innumerable possibilities, affecting everything from entertainment to industrial design, from marketing to healthcare. At Hiredly, we have also seen the potential of Generative AI and have incorporated it into our product features. We believe that AIGC will bring a completely new experience to our Employers and Jobseekers, helping them to work more efficiently, boost their productivity, and achieve personalised experience. Our team spirit is to innovate continuously and we believe that only through persistent efforts can we keep up with the rhythm of the times and create more value for our customers. Generative AI is just one of our steps in this direction. We will continue to seek more innovative technologies and apply them to our products and services to provide our customers with a better user experience. We look forward to exploring more possibilities together in our future journey. --- # 通往未來：從人類學習機器語言，到機器學習人類語言猶記中五那一年，我國推介多媒體超級走廊時期，當年的internet被稱為”資訊高速公路“，我發表了一篇文章，還把它貼在課室的佈告欄上，迫不及待想跟同學們分享一些科技心得。這次又泛起漣漪，熱血沸騰的想要寫一寫，呵呵。身為一位大學畢業後進入科技行業的我，對於那時我們所有人都在學習的機械語言感到驚訝。我們曾經專心致志地學習C、C++等編程語言，使我們能夠與機器對話，並向它們傳達我們的指令。然而，世界的轉變總是出人意料，時代的變化也總是讓我們驚訝不已。現在，我們的目標已經不再是學習機械的語言，而是讓機器學習我們的語言，模仿我們的思維模式。換言之，我們不再僅僅是指揮機器完成我們的指令，而是訓練它們理解我們的意圖，甚至去預測我們的需求。幾年前，我們一直以來的觀點是：先有數據，後有人工智能。然而，現在這種觀點已經過時。Generative AI的出現讓我們有了新的視野，它可以直接用人工智能生成新的數據。這種創新的技術提供了無數的可能性，從娛樂到工業設計，從營銷到醫療，它都能產生重大影響。在Hiredly, 我們看到了生成式AI（Generative AI）的潛力，並且把它融入我們的產品。我們相信，AIGC將為我們的客戶和用戶帶來全新的體驗，讓他們可以更有效地完成工作，提升他們的生產力，並且實現個性化體驗。我們的團隊精神是：不斷創新，與時並進。我們相信，只有持續努力，才能把握住時代的節奏，並為我們的客戶創造更多的價值。Generative AI只是我們的其中一部分，我們將繼續尋找更多的創新技術，並將它們應用到我們的產品和服務中，以提供給我們的客戶更優質的用戶體驗。我們期待著在未來的旅程中與你們一起探索更多可能性。

Building an Enterprise Chatbot

Thu, 02 Jul 2020 08:52:06 GMT

Bot in A Box I observe that some people have this perception that chatbots are actually called a chatbox. Perhaps this is due to the typing box we usually see at the bottom of a chat window. Well, what is a chatbot? In short, it is software that has the capability to interact with a human in a natural unforced way via a conversational communication method, either by text or voice. Some people calls chatbots, virtual assistants. Brief History of Chatbot The method was used in the Turing test in 1967 known as the world’s first chatbot recorded that interacted with a human. The evolution then started quite markedly after 16 years, with more chatbots like ELIZA, JabberWacky, and ALICE being developed. ALICE was my first interaction with a chatbot during my university study on AI. Now we have many modern chatbots which demonstrate the competitive evolution, led by giant tech companies, e.g. IBM’s Watson, Apple Siri, Google Now, Amazon Alexa, Microsoft Cortana and even Facebook Messenger bot. Architecture High level architecture In a high level design, a chatbot must be able to detect human input accurately and also be able to transform back from a digital result to a humanised sentence. I found that this is very practical and useful. This flow can easily be scaled up to include quite a complex conversion of commands. Build from Scratch or Predefined Model? First of all, you need to have a very clear objective to build a chatbot and have complete clarity on the problem that you want to solve. If you want to enhance your business operation by automation and ease any manpower bottleneck, then there are plenty of built in, or predefined models on the market that will meet your needs. You do not need to build from scratch unless you are looking for a highly intelligent, bespoke, and mixed-conversational AI solution. By leveraging on a predefined model chatbot, you can always scale up and improve your chatbot by using the collected data. Each enterprise should make use of your collected data to find any gaps beyond your legacy business. The Trend According to Gartner’s study, chatbot market size is growing rapidly globally. It is estimated that the market will be worth $1.3 billion by year 2024. 47% of organisations will use chatbots for customer care40% of enterprise will deploy virtual assistants. Chatbot is expected to be found in high mobility channels, mobile applications, wearable smart devices, and in-car entertainment systems. In the near future, voice bots will become mainstream. Conclusion Defining KPI measurement success is a key step in every enterprise AI solution. This will keep reminding you “why” you started the chatbot project in the first place. It is important to roll out the chatbot quickly with a minimal viable product. That way you can monitor and always ensure you have the flexibility to enhance your chatbot when required quickly and for minimal cost. Finally do not forget to include an accessible human intervention element in your design. It is a crucial to be able to handle complex conversations that require human escalation to answer the customer’s queries when the answers are unique. That way you make best use of all aspects available resulting in a streamlined in tune service for your customer’s needs. All the best!

AI can do lots, but there are limits

Thu, 25 Jun 2020 08:52:06 GMT

My Impression of AI I had read an article(published in a magazine) during my secondary school time. Unfortunately I cannot recall the title. The story background was about a future city where human and robots were living together in harmony. The robots looked so similar and were capable of doing things just like the humans. Both collaborated in various industries, including politics, education, transportation, etc. The only way you could find out if they were a robot or human was to open up the skull to see whether it housed a circuit board or a human brain. Imagine how strong and perfect AI was perceived in such a community in that city. Not the Terminator-AI yet I believe this is a common misunderstanding on AI, i.e. people often anticipate AI can work like magic, or it is something so advanced like that is often portrayed in the blockbuster movies. If you build a robot, people tend to imagine that it looks and acts like The Terminator type, i.e. a highly intelligent machine that looks like and acts like a human-being. In fact, AI today, generally known as a third wave breakthrough is mainly achieved through a deep learning algorithm. This is applied widely on machine vision and audio. And, I must mention that, this is a fully supervised learning result. Not a mixed Conversational-AI yet Some countries in this region like Malaysia or Singapore are so unique of their rojak(mixed) culture which results in their truly localised language speaking environment. I have seen how good a conversational AI device, e.g. Alexa, Google Home can handle English conversation; yet we need a true rojak AI to appear in our space. Apparently, in my view, a teacher’s job is still quite secure, given that not everything fits nicely into a basic straightforward model. In view of this multilingual teachers cannot be easily replaced by AI in this country. By the way, I remember that probably three years ago there was a mini robot called Robi that claimed it was designed to interact in English, Malay and Mandarin. It would be really helpful to share your experience of interacting with Robi. Not a Creator-AI yet Human emotions included in a machine have always been controversial. You might argue that reinforcement learning could enable a machine to have a self-consciousness capability. Well, there is no denying that AI has enabled a machine to simulate human’s thinking and behaviour. However, in my own personal opinion, this is still very far away from being on a par with the true actual human emotion level. Given this we could notice a low adoption of AI in creativity industries like artists, software programmers, researchers, scientists, etc. All these jobs require a high consumption of human brain cell activity and quite often the end product designed will be bespoke and created for a single specific purpose if a human was producing it. Conclusion Today, the so called intelligence is based on Turing test which aims to solve a problem in a mathematical way. Is strong AI possible? Will a computer really think? Well, in my opinion, we are at the very beginning stages of this journey. There is much of the path still to tread before we can be absolutely certain what is possible. Let’s focus and appreciate what benefits AI can bring humans at each breakthrough stage. This will enable us to strive for the best solutions for the global world.

【人物BAR | ThengHui 张腾辉】AI 新时代，我们的生活

Tue, 12 Nov 2019 08:52:06 GMT

你有想过，AI 功能占领了我们日常生活的多少吗？你有想过，我们已过度的依赖 AI 功能带来的便利？那你可知道什么是 AI？生活中有哪些东西是涵盖了 AI 功能？而 AI （人工智能）带给人类的是生活便利？还是另一种潜在的威胁呢？假设 AI 在未来的 5 年内，大幅度的进化并能够与人类正常生活、互动与交流。我们是否能够和 AI 共处？身处 AI 时代的你，非知不可！这期的《人物BAR》我们邀请到了身为Head of AI 兼 AI 讲师 — 张腾辉先生，与我们解说 AI 与人类这当中的疑虑。

My First Experience with Innov8rs

Tue, 29 Jan 2019 08:52:06 GMT

### Innov8rs @ Bangkok, 23-24 Jan 2019 For the first time, I participated in this innovation conference. The purpose of this conference is to educate the big companies how to face the challenges of modern innovation. Everyone knows that the larger the company or organization, the slower the transition and the lower the success rate of innovation. In the two days of the conference, many domestic and foreign mentors shared their experiences and journey of innovation. In fact, our company has already carried out some of the mentioned methods. It is good to know that they have got validated by the mentors. ### Everything is 4.0 From the era of Internet, to the era of information explosion during my high school time, then followed by the application of big data and artificial intelligence nowadays. Technology is changing fast with each passing day. As a technology practitioner, you must also upgrade yourself to version 4.0. Otherwise you will be out of the game very soon. ### The top management to know what to change To be honest, a company needs to transform, to innovate, to create a new corporate style, but it is always easy to say than do. In this case, according to mentors’ validated experience, we must get the insight and support of the top leadership which will drive the entire company and create innovations. The truly advice would be “Without the awakenings and blessings of the top leaders, do not start this!” Every time we talk about innovation, mindset and culture are the important keys to success. I believe everyone knows that it is impossible for one person to change another person’s behaviors, but everyone has the ability to improve themselves, create their own new values, and re-adjust their ability, speed, thinking, action, etc, until the innovation force is fully realized. In this way, even though it is a century-long family business or a giant organization, you still can sail further to find your blue sea in this challenging waves. ![Participants](img/trip-mate.jpeg) ### Talking about money is not sad When big companies talk about innovation and needs to adopt new technology, they are often cut off by CFOs. Why? The reason is very simple that the treasury masters often cannot measure what hasn't happened yet. For them, the past financial report and the numbers are always the best justification. Do you think employees are assets for a company? There is no doubt that the leader of innovation certainly says that talent is an asset of invaluable treasure! However, for the finance officer, people is not an asset at all, it is a liabilities to the company indeed. Haha. To convince your CFO, you don't just need to calculate the ROI [return = (budget's turnover - cost) x success rate], but also to calculate the growth value at another perspective. For example, AirBnb does not own a room, but it was valued at US$31 billion. In contrast, Hilton has 890,000 rooms, but the market value is only US$23.3 billion. ### The HR has to keep up with the trend Now the personnel department has to be at the cutting edge of innovation too. This is because modern young people are looking out for something like the boss’s entrepreneurial journey, the company’s culture, the company’s interaction with employees, and compensation plan. All these criterias are impacting their stay/leave decision. Masters share that a short-term rewards (for example, upon completed a project or important milestone) sometimes motivate young generation than a legacy practices of year-end bonuses. In addition, personal training programs are far more important than personal accomplishments, because young people are eagerly pursuing on new skills nowadays. Therefore, the seniority can no longer be used to measure employee loyalty, but rather to measure productivity and employee's contribution to the company. ### New collaborationship According to stats projection, from now until year 2040, Gen Y and Gen Z will be the majority in the workplace. This is because many large organizations today have started to explore how to work side by side with startups to increase creativity and achieve a win-win situation. Of course, our company has no exception. Our mobile app is developed in cooperation with startups. You will definitely feel much younger when you’re working with energetic startups. Their efficiency is extremely higher than the usual practice of large institutions. Imagine that a big company is like a dinosaur. Occasionally, if it wants to turn around, it feels that it is too cumbersome. ![Event Speaker](img/event-speaker.jpeg) ### Cultivate intrapreneurs Many small companies have begun to pamper their talents, hence large organisations must know how to retain talents too. The masters said that the company should invest on the employees, instead of reluctantly having employees resignation and letting them to go to competitors. It is better to shape their design thinking mindset, encourage employees to develop and start businesses internally, and then the company will get more returns than just having one or two value proposition of a new products. The company would then have its own ecosystem at the same time too, so that it is not easy for competitors to break through. ### Summary In conclusion, 80% of the success factors that went to innovate depend on the company culture. In addition to the need to clearly understand the goals and visions, everyone must work together to make a difference. Remember, you must not fear of failure. You will need to fall fast and climb fast, then you can succeed quickly. I wish you all that your innovation journey is not lonely because you have us! *[Credits to all the mentors & speakers at Innov8rs]*