Since its invention more than half a century ago, computers have become a part of our daily lives. Many tasks can be reduced to a series of computations and therein lies the reason for the ever increasing pervasiveness of computers. Tasks that need to be executed repeatedly on a large scale are often ripe for automation.
However, automating tasks is traditionally done via programming and this requires people to learn a computer language to communicate and instruct the computer. Learning a new programming language or a new application can be daunting, especially for non-technically savvy folks. That means the majority of us. TagUI, a free Robotics Process Automation (RPA) tool by AI Singapore, hopes to revolutionise automation by getting the computer to understand human languages instead.
For TagUI, bots are built as text expressed in a human language. For example,
type search-bar as hello world
types “hello world’ into a web input search bar, and
clicks the search button on the webpage.
Easy enough to understand, even if you have never programmed before, right?
With such intuitive ease of use, bots can be easily shared with and edited by your teammates. Anyone can access these bots with their favourite text editor, such as Microsoft Word, Notepad++ and more. With TagUI’s new Microsoft Word and Excel Add-ins, users can even create, edit, deploy their RPA bots in MS Word, and define data for batch automation in Excel (link).
With this foundation, it is not surprising that TagUI can already understand commands in over 20 different human languages, bringing down the barrier to automation and democratising the use of RPA in over 80 countries. This would not have been possible without contributions from the community of users. The video below shows how you can use the four official languages of Singapore – English, Chinese, Malay and Tamil – to automate the task of getting a list of stock prices from Google.
Interested to know more? Get started with TagUI here!
Earlier this month, I had the opportunity to share in a webinar about AI Clinic and AI Discovery, part of the comprehensive programme in AI Singapore (AISG®) to help organisations accelerate their AI adoption. In this article, I will elaborate more on what goes on in an AI Discovery workshop where AI use cases are ideated and prioritised.
AI prototyping, unlike software prototyping, requires preparing data and experimenting with algorithms to determine the feasibility of use cases. Both cost time and money. Organisations must ideate the right use cases to pursue to reduce wasted efforts and increase the probability of AI adoption.
As an Assistant Head at AI Singapore, I have witnessed common mistakes that business leaders made when identifying AI use cases. These mistakes resulted in the development of unfeasible or unimpactful AI use cases that continue to haunt the organisation and put a dent to future interest in AI. These could have been minimised.
Common Mistakes of Ideating AI Use Cases
1. Didn’t minimise groupthink
Business leaders will typically gather representatives from different departments to brainstorm on potential AI use cases. The brainstorm session usually starts by either business leaders sharing their thoughts, employees presenting their ideas, or with an informal roundtable discussion.
“The first project that the CEO suggested is not the right one to invest in.”
– Andrew Ng, CEO and Founder of Landing AI, during Amazon re:MARS 2019
All of these approaches have serious flaws: groupthink usually happens after senior management have shared their thoughts; good ideas might have already been filtered out before presentation to management; extroverts will dominate discussions. All this will result in a list of low-risk and ‘common-sense’ ideas that fail to harness the full potential of AI.
2. Didn’t align teams
Deploying AI solutions is beyond simply installing new software. AI, given its capabilities, will often require organisations to adjust their business processes and employees to adjust their workflows to harness the full potential of it. In other words, successful AI model deployment would require aligned teams.
Business leaders, however, often overlook team alignments during ideation. This results in a lack of support and adoption of AI model as teams with different priorities will not invest time and effort to adjust their existing processes. The AI solution will die a natural death.
3. Didn’t engage facilitator who has AI expertise
I have seen companies selecting use cases that are not AI-related after their internal ‘AI ideation sessions’. The root cause? No one in the session had expertise in AI. They were unable to decipher whether the selected use cases are possible for AI development. It is unproductive as the participants have to redo the ideation exercise to identify better and relevant AI use cases.
For instance, one of the companies I worked with chose ‘using AI to ensure customers are onsite to collect delivery’. This use case is more of a business process problem than an AI use case. It could have been archived earlier during the ideation workshop to focus everyone’s attention on AI-related use cases.
What is Design Sprint and How It Can Help?
Google Ventures developed Design Sprint to ideate and test use cases in a structured approach. It is designed specifically to addresses the common mistakes of brainstorming mentioned. The original process, however, is not optimised for ideating AI use cases.
Based on my experience working with various companies, I have adapted it for ideating AI use cases by making one major modification: the facilitator will filter out non-AI use cases when participants are voting on challenges to focus on. In addition, the first part of the workshop will end with ‘search for inspiration’ instead of ‘target mapping’ as described in the original Design Sprint.
The best way to understand how the modified ideation process works is through a case study. The case study below is from an actual session that I conducted; the company name and sensitive information have been masked for confidentiality purpose.
BETTER BRAIN is an institute of higher learning. Their CEO is interested in using AI in their organisation. However, the CEO is unsure where to start. He decided to engage AISG to facilitate an ideation workshop for his team.
Step 0: Preparation work
The facilitator will discuss with the project sponsor or the most senior person attending the workshop to identify a suitable theme for the workshop. This puts a soft boundary around exploration to keep the discussion within a defined space.
The facilitator will also advise who should be attending the workshop. The best is 4–7 participants from different departments/functions that are related to the theme of the workshop.
The CEO of BETTER BRAIN has decided to gather employees from marketing, IT, and programme to attend the workshop. He has also decided on the theme: improve learners’ experience and outcome.
Step 1: Participants sharing and converting challenges into ‘How Might We’ (HMW)
Each participant takes turn to share their business challenges. The rest of the participants, including the facilitator, will convert the challenges shared into opportunity areas. The questions will be written in ‘How might we…’ (HMW) format.
The purpose is to generate potential ideas to solve business challenges shared. As each participant has a different perspective, there will be different ideas generated for the same business challenges. There is no discussion among participants to minimise groupthink and prevent extroverts from dominating the discussion.
“We use the How Might We format because it suggests that a solution is possible and because they offer you the chance to answer them in a variety of ways. A properly framed How Might We doesn’t suggest a particular solution, but gives you the perfect frame for innovative thinking.”
In the example below, after the Programme Manager from BETTER BRAIN shared her business challenge, the rest convert it into questions in HMW format:
Step 2: Categorise and vote on challenges
Facilitator and participants will identify common themes across the HMWs and categorise them accordingly. Each participant is then given two votes to decide which are the challenges worth pursuing. There is no discussion among the participants.
The categorising of challenges is to facilitate voting; the voting on challenges is to identify important and relevant challenges that the team wants to pursue.
The illustration below shows the result of voting for BETTER BRAIN:
Step 3: Filter challenges
Challenges with less than two votes are removed. The facilitator will also remove challenges that are non-AI related and explain to participants why they are non-AI related; this is one of the modifications made to the original process.
After this step, all the remaining use cases are potential candidates for AI solutions development.
Step 4: Define goal
Each participant writes a goal in “In two years’ time, we will…” format. Participants then vote to decide which is the goal worth pursuing; everyone has one vote.
The purpose of this exercise is for participants to envision from their perspectives, what the company or business unit will look like if all the challenges identified are solved.
For BETTER BRAIN, the participants have identified that their key goal in two years’ time is to make use of data to guide students on selecting suitable courses.
Step 5: Identify blockers
Each participant will list down 2–3 potential blockers that could stop them from reaching their top voted goal. The blockers should be written in “Can we…” format. Each participant is then given 3 votes to select the most relevant blockers to the goal identified.
The purpose is to highlight potential pitfalls that could derail the AI project or prevent the team from achieving their goal. This allows management to preemptively address them to increase the likelihood of AI project success.
For BETTER BRAIN, the participants have identified the top three blockers that could stop them from achieving their goal:
Step 6: Search for inspirations
Participants are given 15 minutes to search for 1–2 inspirations to solve similar challenges identified and note down the key ideas of each inspiration. These inspirations need not be from the same industry.
By the time participants reached this section, they would have initial ideas on how potential solutions might look like. The challenge with sharing ideas is that they might be hard to explain and people have different interpretations of the same abstract explanations.
Therefore, the purpose of this section is to let participants manifest their ideas in concrete form and share the key applications of their ideas.
At the end of the Mapping process, the participants will have the following:
List of identified opportunity areas and potential solutions.
The goal that the participants envisioned the AI use cases would help achieve and potential blockers that might derail the AI project.
Alignment among participants, through the series of voting sessions, on key issues to tackle.
The management could consider these inputs when deciding which AI use cases they should be developing.
Identifying feasible and impactful AI use cases is challenging. The common approaches used by organisations to identify AI use cases are unsuitable for the ambiguous and data-dependent nature of AI. AISG has adapted the Design Sprint into a structured and proven approach to ideate AI use cases. If you wish to find out more, you can reach us at email@example.com for further information.
There are many technical tutorials out there, but few offer the hands-on experience needed to address real-world problems. And that, according to engineer Derek Chia, is one of the key differentiators of AISG’s AI Apprenticeship Programme (AIAP)®.
Derek graduated with a Bachelor of Computing in Information Systems and was a cyber security consultant with EY Singapore for more than two years when he decided to join the pioneering batch of AIAP apprentices in May 2018.
“Artificial intelligence (AI) and machine learning (ML) is a rapidly growing field with impactful real-world applications. I wanted to advance and expand my skillsets in this area,” he said.
The self-directed learning style of AIAP allowed the apprentices to take a deep dive into various technical areas such as programming, algorithms and tools, and to learn together as a group. Looking back at the experience, he said, “The area of AI/ML is ever evolving, and the ability to learn how to learn is a skill I cannot emphasise enough.”
The programme also provided Derek with the opportunity to put what he had learnt to good use. For example, as part of AISG’s 100Experiments (100E) programme, and with the support of experienced mentors, the apprentices got to work alongside AI engineers and project managers to help industries solve business problems using AI.
For his 100E project, Derek worked on the problem of wrong delivery addresses for e-commerce orders. “Addresses from the e-commerce platform may be spelt wrongly or mis-constructed, and so the goods get delivered to the wrong location based on what the mapping engine (Google Maps) tells us,” he explained.
To address this, the team carried out entity extraction and text correction on addresses captured in the transport management software. “Based on our experiments, we found that correcting the addresses using our model before querying the mapping engine improved accuracy and helped to speed up processing time,” he said.
It also reduced costs by minimising the need to query the mapping engine again, which was being charged on a per-API call basis, and by reducing the need for re-delivery.
Among the many useful skills Derek picked up whilst working on his 100E project was the ability to communicate with business users and discover the crux of a problem before searching for an appropriate solution. “AI/ML is no magic pill, and identifying the right problem to solve is one of the important things that I learnt through the 100E experience,” he said.
After AIAP, Derek started a new career as an engineer with the Defence Science and Technology Agency (DSTA) – a job which he landed even before he graduated from the programme.
As he looked back three years on, the AIAP experience prepared him well for this new role, he said.
“AIAP empowered us with new skills and opened my mind to new possibilities.”
His current job involves applying AI/ML techniques in multiple areas such as natural language processing (NLP) and speech recognition. As he immerses himself in this field, he is constantly being motivated by the application of cutting-edge research to real-world use-cases and seeing it solve problems in ways that were never possible before.
The discipline of self-directed learning, honed over the course of his apprenticeship, continues to stand him in good stead, allowing him to keep abreast of the latest developments in technology.
“Having a growth mindset, being disciplined and taking responsibility for your learning outcome will take you through AIAP and beyond,” he said.
The primary role of an information retrieval system (IR) is to retrieve a set of relevant documents given a query. For a machine learning based IR to be effective, it needs to be trained on sufficient training data and its performance needs to be monitored on a regular basis to avoid a “data drift” situation.
“Data drift” happens when any of the following scenarios occurs:
Users start to phrase their queries differently.
Users post queries that the IR is not well-equipped to answer.
The knowledge base content underlying the IR system is no longer relevant and needs to be updated (eg. HR policy before and after COVID-19).
Beagle was developed for IRs to be monitored in an intuitive and simplified manner and to collect more training data. Before delving deeper into the inner workings of Beagle, we will further illustrate the concept of “data drift” in the context of information retrieval.
Imagine that you have an IR that is deployed within your organization to answer COVID-19 related questions and one of your users posted a question with regard to the circuit breaker measures implemented by the Singapore Government. Unfortunately, the IR was not trained to recognize that the term “circuit breaker” in the context of COVID-19 and without adequate measures in place to look out for such occurrences, the poor performance of the model might be left unnoticed.
User: “What are the current circuit breaker measures?”
What the IR returns:
“A circuit breaker is an automatically operated electrical switch designed to …”
What the user is actually looking for:
“The 2020 Singapore circuit breaker measures, abbreviated as CB …”
Beagle in Action
A maintainer, who is usually an internal staff of the organization deploying the IR, will leverage on Beagle to collect more training data, finetune and deploy a new IR model. On the other hand, end users will interact with the IR by posting their queries to it. In this section, we explain and illustrate the interactions between Beagle and the IR using the architecture diagram below.
1. User sends query and receives responses
End users post their queries to the IR and in return receive a set of relevant responses.
2. User provides feedback (optional)
End users are able to provide the maintainer with feedback on how the IR is doing by selecting the responses that are relevant to them.
3. Maintainer verifies feedback
There might be cases where the end users’ annotations were incorrect. As such, Beagle has a similar interface for the maintainer to verify the feedback given by the users. The data collected will be especially useful for companies who do not have the capabilities to annotate their own training data.
As the IR returns a fixed number of responses (eg. 5 responses), it is possible for some ground truth responses to not appear in the list of responses returned to the end user (false negatives). The maintainer is able to make corrections for such errors by searching for these responses and annotating them as correct responses. Once the end user’s annotations have been verified and corrected, they will be saved and used for downstream model training purposes.
4. Maintainer has the option to upload labeled data
Alternatively, for companies who are able to annotate their own training data, Beagle allows the maintainer to upload the labeled data in the form of a “.csv” file.
5. Maintainer evaluates model against new batches of data
The data that is being collected over time are separated into batches of fixed size (eg. 100 data points) and are evaluated against the currently deployed model. Each data batch is split into train, validation and test sets.
For information retrieval tasks, Mean Reciprocal Rank (MRR) score is a typical metric used for evaluation. As an example, the graph below illustrates the performance of the currently deployed model across 9 batches of data. Each of the data batches has two MRR scores (“Batch score and “Accumulated test score”) tagged to it. For clarity, we have listed down the data points that were used to calculate each score for batch #8 below.
Used to Compute Batch Score for Batch #8?
Used to Compute Accumulated Score for Batch #8?
Yes – Only test set
Yes – Only test set
Yes – Only test set
Yes – Only test set
Yes – Only test set
Yes – Only test set
Yes – Only test set
Yes – Only test set
6. Maintainer retrains a new model
If required (e.g. when data drift has severely affected the accuracy), the maintainer is able to retrain a new set of model weights for the IR with the data collected so far.
7. Maintainer deploys a new model
After retraining a new set of model weights, the maintainer has the option to deploy the new model. Additionally, the maintainer has the option to revert to an older model version if he/she finds that the performance of the currently deployed model is unsatisfactory.
The maintainer continues to monitor the performance of the model as new data is collected.
Our motivations for building Beagle for IRs are aligned with the Continuous Delivery For Machine Learning end-to-end process illustrated below. The data collection and performance monitoring features of Beagle tie in closely with the “Monitoring” and “Model Building” pipelines. We are continuously improving the functionalities of Beagle and are looking to generalize the features of Beagle to other machine learning use cases. Do drop us an email if you have any questions or if you would like to request for a demo.
Last week, AI Singapore (AISG®) held a webinar to share about the AI Clinic and AI Discovery initiatives which have been designed to help organisations with their AI transformation journeys. As a bonus, we also had one of our collaboration partners, uParcel, join in a fireside chat to talk about how they had benefited from their participation.
For those who missed the session, the full video and the bookmarked segments are available below. You can jump to the parts which interest you by clicking on them.
Interested to start your own AI transformation journey? You can contact the AI Advisory team at firstname.lastname@example.org for further information.
1. Introduction to AI Singapore
If you are new to what AISG does in general, this segment provides an overview of all our programmes and the work we have done since our establishment in 2017. ( Jump to video segment)
2. Challenges of Adopting AI
After having implemented AI solutions across the nation for almost four years, AISG has accumulated a lot of experience on the various challenges to AI adoption. The AI Clinic and AI Discovery have been designed to meet these challenges. ( Jump to video segment)
3. Introduction to AI Clinic
For companies new to AI, the AI Clinic has been designed to provide relevant information on AI use cases by industry or function. ( Jump to video segment)
4. Introduction to AI Discovery
For companies ready for a one-to-one engagement with AISG, the AI Discovery offers close consultation on AI use case ideation and prioritisation. ( Jump to video segment)
5. Fireside Chat With uParcel COO William Ng
William Ng, the COO of uParcel, a leading on-demand 24/7 courier service in Singapore, shares about his experience working with AISG to build an intelligent aggregate routing solution that has boosted the company’s productivity by 20%. ( Jump to video segment)
At the start of the year, I had a chat with Kevin, our head of the AI Advisory team here at AI Singapore (AISG®). We talked about our personal experiences with career transitions into AI. In this conversation, he shares more about his work at AISG helping companies, large and small, to chart out their own journeys with AI.
Below is a transcript of the conversation [*]. If you are keen to know more about the programmes mentioned (including the 100E), you can contact the AI Advisory team at email@example.com for further information.
Basil : Hi Kevin. Great to have you here again.
Kevin : Hi Basil. Good to be here again.
Basil : So, previously we talked about mid-career transitions for mature PMETs and today you find yourself leading the AI Advisory Team in AI Singapore. Tell us about your role and the mission of the team.
Kevin : Sure. So, currently I’m the head of AI Advisory in AI Singapore. The main responsibility of my team is, in a nutshell, to help Singapore companies, both large companies and small ones to adopt AI. As you know, AI is fast becoming critical to the competitiveness of businesses in many different industries. In some industries, AI is no longer a nice-to-have, but rather a critical tool that determines the survivability of the business. That’s why I think the Singapore government has invested heavily to help local companies, as well as Singaporeans, to upgrade themselves to be AI-ready, so that our economy can be competitive in the coming era. Therefore, our job is to catalyze this transformation. We run different digital transformation programmes for large corporates as well as SMEs, even start-ups, so that they can run faster in the adoption of AI technologies.
Basil : Yes, I think this is certainly an important component in the next stage of our nation’s development. So, as you mentioned, our economy has become more diversified since independence. Different organisations would therefore have different starting points on their journeys to adopt AI. Could you elaborate on how you address such a differing range of needs?
Kevin : Certainly. Yes, companies have very different maturities and readiness. They are at different stages along their digital transformation journeys. On the one end of the spectrum, you see companies that are interested in AI, but they don’t really know much beyond that. For these companies, our goal is to help them understand what is AI and how does AI help their business. Unfortunately, there are lots of confusion and misinformation out there. So for these companies that have just started on their digital transformation, it could be quite daunting. We try to help them with things like AI Clinics, which are workshops targeting specific industries. We invite people from that particularly industry to the AI Clinics. These participants are typically business owners, business unit heads, decision makers from a specific vertical industry. In that workshop, we share with them success stories and AI use cases pertaining to their industry. We go beyond theory. We focus on practical aspects of AI in business. And important to note that our team members comprise professionals who are technically trained in AI with many years of commercial experience. For example, I myself am a certified AI engineer, but I have also spent over twenty-five years managing businesses. Therefore, we are able to talk to business owners in languages that they can understand and also help translate the benefits of AI in a way that makes sense to them and through these AI Clinics, we hope to trigger their curiosity and create the interest in them to explore AI in their own companies. So, the objective is really to equip the participants with enough knowledge so that they can start the conversation within their organisations about AI.
If they get to a level of interest that they want to explore potential ideas, then at that time they can come back to us. When they do, we have another programme called the AI Discovery, which is a one-to-one consulting engagement to help companies to ideate and also to prioritise those ideas into actionable next steps. In the AI Discovery programme, companies will share with us their business priorities and their goals and then we will help them look at which ones of those are solvable by AI. As AI Discovery is often conducted under NDA (non-disclosure agreement), companies usually will share with us their actual data sets. With the data sets, we are able to help them look through, to see whether they have enough data and the data is in the right format and whether they are appropriate for the AI projects they want to embark on. At the end of the AI Discovery, the output is a consulting report with specific recommendations. These recommendations could be, let’s say, an AI project if they are ready to do that or just suggestions on the specifications and the project goals. But sometimes, we might also find that the companies do not have enough data to support the project and in those kind of cases we then will recommend to them what type of data they would need to collect and how to go about doing that. And then after that, for those companies that are ready to do an AI project, we will take one step further to recommend to them various ways of doing that. For example, the most obvious one is they could explore the 100 Experiments (100E) programme in AI Singapore. The 100E is a heavily subsidised programme for companies that are embarking on their first AI project and we focus on projects that require technologies that aren’t readily available in the market. Alternatively, they could also make use of something called Bricks. In AI Singapore, we develop reusable AI technology components which we license to local companies either at no cost or at very low cost so that we can help them speed up their AI project or to reduce their cost. Lastly, if we find that the best alternative is to make use of commercially available products or services, we would also advise our customers to do so. As AI Singapore isn’t a commercial entity, our end goal isn’t really about selling a service or a product to companies, but rather to give them the best advice on how they can go about doing their AI project. Therefore, we will try to help our customers in ways that are best for them.
Basil : I think the fact that AI Singapore isn’t so-called commission-based is a unique reason to engage us. Maybe let’s take a step back. What are the kinds of opportunities you see as impetus for companies to adopt AI?
Kevin : That’s a great question. Many companies do see that the current environment around them is fast changing and business owners whom we have spoken to told us consistently they’re looking for opportunities either to increase their sales or to reduce their cost, especially after the COVID pandemic. These are the two things that are at the top of the minds of business owners, whether they’re big companies or SMEs. These are the things that the bosses are most worried about. The ability to do so is critical to the survival of their business. They know that business-as-usual after COVID is not an option. So they want us to help them understand how they could leverage AI to achieve either or both of them. Therefore, I think scoping the opportunities of AI from the perspective of either increasing sales or reducing cost is the easiest way for business owners to understand.
Let me give you an example. AI is able to help companies understand their customers better, like which customers are likely to purchase what products from you in the future and which customers are potentially at risk of churn, meaning stop buying products from you in the future. So if you know that, then you can take appropriate interventions to either stop the churn or to sell more products. In the past, such knowledge might reside in the heads of very experienced employees or just within the sales organisation. The problem is, such knowledge is hard to retain. When the experienced employees leave, they take the knowledge away with them. It is also very difficult for companies to institutionalise and share such insights throughout the organisation even if they wanted to. By using AI, we can extract such insights out from historical data in a much more systematic and data-driven way. With the help of AI, companies could potentially find out more about their customers in a way that they couldn’t otherwise. There are a lot of times after doing the AI project they find things about their customers they didn’t know in the past. Therefore, such insights can then be shared broadly across their organisations helping decision makers to make better decisions, as well as more timely decisions. So, such things like demand forecasts, customer recommendations, sales prediction etc are very popular requests among companies in Singapore. AI can definitely help our companies to do these things more effectively than the traditional way and hence increase their sales opportunities. Likewise, there are many opportunities for cost reduction as well. You know, people talk about automation, that is obvious. That can be done by digitalisation, but AI can take that one step further. For example, AI can help businesses do better product quality checks through computer vision. AI can also process customer requests through NLP (natural language processing). NLP is a branch of machine learning that has improved tremendously over just the past few years. This technology is now able to understand human language, whether it is text or verbal. So with that, companies can actually automate many mundane tasks of responding to customers in a much quicker way, but at the same time reduce the cost of doing. These are some of the examples that we see in the market in terms of the potential of increasing sales or reducing cost.
Basil : What about at the individual level? What does it mean for the individual in such a AI-pervasive future?
Kevin : Yes, that is an important topic. We read a lot in the media and sometimes people worry about AI replacing jobs. Actually, I think that is a misconception that is unfortunately perpetuated by popular media. In reality, AI can help automate tasks but often not the entire job. There are some jobs that can be totally automated away, but that’s not very common. On the other hand, the tasks that AI can help to automate and do better are typically those that are mundane and repetitive. AI can actually make the job more interesting for humans by letting them focus on tasks that require creativity and human empathy, for example. So the reality is, AI will not totally replace humans, but rather those who learn how to use AI as a tool will become more competitive against those who don’t. Therefore, I think it is important for everyone to understand what AI really is and to learn how to make use of AI as a tool instead of fearing AI or thinking AI is some kind of a magical wand that can somehow solve all kinds of problems automatically. Not everyone needs to become an AI engineer, but it’s critical that everyone takes the initiative to pick up the skills to use AI tools. In the future, AI tools will become as common as using Excel or using online tools in our daily lives today to do our business. So for companies, I think the important next step is for senior management to internalise how digitalisation and AI are strategically important to them and then to start the transformation process for their companies. I think that’s the most important first step for companies to do. This transformation requires a big mindset change. A change from doing business solely through experience and intuition to one that leverages the power of data and AI. You know, as people say, AI is a foundational technology. That means it is something that transforms how businesses are done throughout the company. It cannot be viewed as something just for the IT department to worry about. The attitude cannot be, let’s provide some funding, let the technical department try out some AI projects and life goes on as usual. That definitely won’t work. Instead, work processes need to be redesigned in such a way that AI can be incorporated into all the appropriate areas within the organisation. Therefore, it has to start from the top.
Basil : Right, in any kind of change, especially at the organisational level, there will be resistance. Could you elaborate on how to overcome such roadblocks on the journey to AI readiness and adoption?
Kevin : Yes, I think the roadblock to adoption is really the resistance to change. Many companies, when you talked to them, they do recognise the need to change, but they fear the uncertainties that the change might bring. As I see it, some companies are hoping that they can somehow revamp their business or acquire new capabilities by doing things that they are already familiar with instead of embarking on things that are out of their comfort zone. So with regards to AI, I think that is due to the lack of proper understanding of what AI really is. Many view AI as something that is very technical and very complex. To some extent that might be true, but that shouldn’t be a roadblock. Let me take an example. Many businesses use machines that are very high tech and very complex too, and they probably don’t fully understand how these machines work under the hood, but that’s not a problem. What do they do? They retrain their employees to work with these machines, revamp the business processes to incorporate these high tech machines. Over time, the decision makers will also familiarise themselves. They acquire enough knowledge about these technologies to be effective. So, same thing with AI. In the beginning, it may look daunting, it may look complex, but it isn’t. It doesn’t need to be. It is nothing but a tool. Companies need to get started, start slow and learn how to make use of data and AI, and take their business to the next level. AI Singapore is here to help. We have programmes for both individuals as well as companies to help them along their digital transformation journey.
Basil : So the message is clear. Since being set up almost four years ago, AI Singapore has developed a whole suite of programmes and initiatives for both individuals and companies to integrate AI into their lives and operations. Well, thanks for sharing with us today, Kevin.
Kevin : Glad to be here again.
[*] This conversation was transcribed using Speech Lab. The transcript has been edited for length and clarity.