Building AI workforce

Feeling under the weather, so sharing a podcast by Andrew Ng on building an AI workforce.

A great 24 minutes podcast! Listen if you are interested with this topic.

When WhatsApp, Instagram and Facebook Down at Once

Not sure what’s the main cause but one thing they have in common is Mark Zuckerberg as the major shareholder. Surely this calls for a break up. Or they cannot at least share the same server.

I can live without Instagram and Facebook. But I don’t think I can live without WhatsApp. It was down for more than an hour! It is the main mode of communication, sometimes ppl even response faster through WhatsApp than calls.

I checked my Twitter and saw this:

Thought that was funny. I actually did that too. I’m sure many others did the same.

AI ethics

It’s been the buzzword lately especially among the US tech community. The big events are (1) Google’s AI ethics board has been cancelled, only after 10 days of existence and (2) Facebook’s faced another round of backlash on its biased algorithm (what’s new for Facebook) and being sued by the US Department of Housing and Urban Development that Facebook’s ads discriminate by race, gender and religion.

I used to not like the word ‘ethics’. I think it’s too subjective if you put laws around it. It should be everyone’s responsible to act in ethical manner. But I guess I was naive back then. It’s really not that simple and straightforward. In finance, there’s a reason why ‘ethics’ is such a big component of CFA curriculum which you need to pass. Asset managers want to filter out companies that do not comply with their ESG (Environment, Social and Governance) requirements. Governance is where your ethics play an important role.

Now not just finance. In the world of AI, industry leaders, VCs, regulators, tech employees, universities, essentially almost all parts of life as AI’s adoption increases are paying a lot of attention to this subject. I don’t think it’s a choice, it’s given. We can’t afford to ignore it. Yesterday, Fred Wilson also encouraged everyone to pay attention to AI and the impact on society, in his blog post.

It’s definitely an area that I’m interested in. Will talk more about it some other time.

Implementing AI pilot projects in large organisation

Earlier this week, MIT Technology Review hosted its annual event on all things artificial intelligence (“AI”). It’s my first time following the event, albeit virtually. You can checkout #emtechdigital on Twitter to know what’s being discussed, presented, debated on AI.

I am particularly drawn at Andrew Ng’s session. When it comes to AI, Andrew Ng’s name is not new to us. An MIT graduate, a professor at Stanford, co-founded Coursera, Google Brain, former VP and Chief Scientist of Baidu (Google of China) and all-time AI activist.

According to Andrew, “if you are thinking about embracing AI: just jump in”.

Here’s a screenshot of his advice on “how to jump in”.


I could relate to this, having done a related pilot project, albeit the beginning or rather foundation of AI. If you are in a large traditional company, not only you need the support and buy in from the CEO, but also across the board. Executing an AI-related project involves a continuous or up-hill battle with various departments such as finance, marketing, sales, IT, legal and compliance and it can be very daunting to argue your way through if your project is not a priority of the organisation, especially if it is just a pilot project. People don’t have the visibility of the success of the project, hence they are afraid to commit in terms of time, cost and resources. But that’s the nature of AI-related projects. It needs to be done multiple times – test and learn spirit, to even spot the quick wins.

Hence, time, cost (i.e. money/budget) and resources are the three essential ingredients to successfully implement AI pilot projects. (1) You need time to allow the test and learn series, from collecting, mining, analysing, training and interpreting data to the legality aspects of it. (2) It is costly as you need to pay for the computing power (software, cloud, data) and most importantly to pay for the right people to deliver the project (data engineers, data scientist are very expensive), hence you need to make the budget available (CFOs need to understand this need). (3) You need a team, dedicated resources and the right mindset. The people you hire to execute, must have the agile mindset and patience as along the way, they need to educate the people they have to deal with from various departments to get what they want.

Sometimes even the high-level executives don’t understand what we do. But that’s fine, we are all learning anyway, these are all new things to us. If you are more advanced than them, create the awareness and educate them. It’s a long haul journey. 

Machine learning and AI in flowcharts

If you are interested to understand what is machine learning and AI in the simplest form, I would encourage you to read the articles below, published on MIT Technology Review.

  1. Is this AI?
  2. What is machine learning?

You might need to enter your email address to read the full length but it’s free, so you should just subscribe.

The flowcharts in the articles are very useful for you to decipher what the heck people especially experts are talking about when it comes to machine learning and AI.

Here’s the screenshots:

The articles are written by Karen Hao, whom I mentioned in my post before. She writes very well and I have just subscribed to her newsletter, the Algorithm. You should too!

When is the right time to automate?

Just had an interesting conversation with a long time friend/former colleague (who left the corporate life to focus on her startup) about choosing automation/digitalisation at work via upgrading to systems/platforms/softwares however you want to classify it or as simple as using formula/macro on excel as opposed to doing things manually such as recording customers data, finance or portfolio information, checking inventories, recording claims, etc

Doing all these things manually can be very daunting especially if it is repetitive and looks trivial or worse, menial. It’s a painful process but most of the time, we complain and do nothing about it. We would rather endure the pain because the time required to change or automate will be more than just spending another hour to manually do it. But what we don’t realise is the cumulative hours that we spend doing it and the rate of time spent increases as we grow bigger and more complex. This reduces your productivity and increases inefficiency at work that may not only affect your team members but also other internal stakeholders.

Even not all startups which deliver their products/services online or through apps don’t bother fixing the problem, at least not at the beginning simply because it can be more costly (in terms of time/money) to adopt automation. There is obviously no question as to whether you should automate. The US and China not only are chasing each other testing and learning repeatedly to create and adopt innovative solutions at work but also trying to solve the “future of work” when automation led by AI makes some of these jobs redundant.

But the question is, when is the right time to adopt automation?

I guess there is no straightforward answer to that. In every organization, there will be a time when you are at an inflection point where you have to either pivot or grow/innovate your business. At that time, if you still don’t invest the time and effort to simplify your work or make your process more efficient, you will definitely struggle in the future.  Sometimes the solution is as simple as creating a microsoft excel template with index/vlookup formulas with proper unique identification. That is also an automation without any sophisticated technology. Once you outsource the menial work to machines/systems/softwares, you can carve out your time to do more meaningful work. That will not only help yourself but also your team and the whole organisation. 


In search for more data and better technology

Yesterday, Larry Fink, the CEO of BlackRock announced that BlackRock is going to acquire eFront, a risk management platform called  for USD1.3bn. I gasped. Late last year,  State Street announced that it will acquire Charles River, eFront’s competitor, for USD2.6bn. I think the acquisition of the latter has been completed. In 2017, GIC invested 30% stake in Mergermarket (now known as Acuris), a core provider of data, research and analysis particularly on M&A.

What does this tell you? What do they have in common? What are the motives of these asset managers?

All longing for more data and better technology/systems to provide more sophisticated solutions for the clients or stakeholders. This is definitely a growing trend and I only mentioned a few, albeit the major ones. BlackRock and State Street are already trillion dollar companies (by AUM) and yet they want to make sure they continue to be at the fore front, as AI, computers and machines take over our lives. Well it makes sense, they are big and they have a lot of cash that they can just buy a company that has been collecting data and selling platforms/softwares to clients in the financial industry and by BlackRock and State Street acquiring eFront and Charles River, they are both going to have access to all clients data. That’s the main beauty of the acquisitions.

This also tells you that as an employee, you have to learn how to use these systems and be able to swim across the ocean of data to provide valuable insights in a timely manner. And that is a challenge. In Malaysia, let’s not talk about asset managers acquiring platforms/systems, we tend to just outsource and sign a contract with the vendors. eFront and Charles River are both names that are not new to me. They are vendors that I have dealt with, either at proposal levels or execution levels.

Malaysia has still a long way to go. If you keep outsourcing, you will rely mostly on the external capabilities which is fine at the beginning but over time, our capability cannot be just limited to storing the data in the systems and then generate output. We need to learn how to control the systems/models so that we can customize according to our needs, mostly to make our daily work more efficient.

It’s not just the technical capability that we need to address, but it is also the culture and mindset. Employees are always bogged down with day-to-day work that they do not carve out time to innovate – improve productivity and learn the new way to do things. This needs to be addressed, otherwise we will be left behind. And this is applicable to all other industries. When you have proper systems, you are able to play with your data in timely manner and generate insights to make better decisions.

Quotes from Invest Malaysia 2019

Some quotes from yesterday’s event Invest Malaysia 2019 that are worth remembering:

  • …embrace digital connectivity, use of data analytics and explore opportunities to build digital businesses [Tun M]
  • The rise and fall of nations depend heavily on the strengths of its institutions [Tun M]
  • Hard work, honesty, and a sense of shame at failure will spur us to give our best [Tun M]
  • Pay your workers better if you make more profits [Tun M]
  • Shared prosperity means that everyone benefits and such a situation is conducive for growth and stability [Tun M]
  • In ensuring the sustainability of our economy, we must increase productivity. This depends on quality workforce. Quality workforce depends on quality education. We want our human capital that is e-Ready and e-fit [Tun M]
  • …short term remedies are needed for the long haul [Tun M]
  • You don’t become unfriendly with big market (such as China) [Tun M]
  • Money saved must be invested in good assets that will appreciate in value such as gold, land, buildings [Tun M]
  • With the decreasing purchasing power today, your savings will dwindle faster if it isn’t invested [Tun M]
  • When you reach retirement age, you keep on working [Tun M]
  • A clean government will always outperform a corrupt government [LGE]
  • Need companies to be better, faster, cheaper in delivering great customer experience which are enabled by technology [Surina, CEO of MDEC]
  • Take a look at what you have that is unique to you (or your country) and develop from there [Ho Kwon Ping, Banyan Tree]

How do we promote AI without limiting it too much…

I stumbled upon this article as the headline was just too attractive to me.


I thought this was an interesting study if you want to understand where AI is headed next and the origins of it in the earlier years. The report is done by Karen Hao, an AI reporter at MIT Technology Review. The important to note here is the effort done to perform this study. According to her tweet, she actually downloaded all 16,625 AI papers abstract from an open research database, arXiv. So the findings are really solid, in my point of view.

The study found 3 major trends, as highlighted in the report:

  1. A shift toward machine learning during the late 1990s and early 2000s (from knowledge-based systems)
  2. A rise in the popularity of neural networks beginning in the early 2010s
  3. Growth in reinforcement learning the past few years

The point here is that, in the researchers world, the term AI has existed much earlier since 1950s. The research work certainly has evolved but the application of it has not. Sure, there are thousands of companies trying to use AI to solve problems but the environment does not allow them to foster fast enough with all the backlash received due to tighter regulations, cyber-security issues, data privacy, etc.

A case in point, a report by NBC recently revealed that “people’s online photos are being used without consent to train face recognition AI”.  And it’s being done by IBM.You can read more about it here. So one of your pictures could have been used. This will only make people angry, although may feel ok about it. I’m neutral. To them and other industry insiders, they needed the data.

So where and how do we draw the line here?

Avoid break up; have full control of your own data

So there’s been a lot of noise on the social media platforms, news, about Elizabeth Warren’s (the democrat candidate) bold and controversial plan to break up the big tech companies – Facebook, Google and Amazon. You can read more about it here and here. She’s evangelizing her plan with the hashtag #breakupbigtech and it has went viral. Here’s a screenshot of her tweet.


Elon Musk’s response to this cracks me up.


Things have gone more wild a few hours ago when Facebook removed Warren’s ad because of a policy violation against using Facebook’s logo but apparently Facebook has restored it “in the interest of allowing robust debate”. You can read more about it here. And then Warren tweeted this:


It’s funny to see/read public’s mindshare and reactions on this, including the authorities and big shots. I saw another tweet saying we should break up the federal government. The word “break up” is being overly used!

Anyway, jokes aside, 1st thing that came into my mind when I read about this was:


As it is, we already have many multiple account IDs when we sign up to all sorts of platforms, software, applications. Yes, for some you can just login with your google ID or facebook ID but there are others that are not inter-related to Google and Facebook that require different logins for example my Malaysian Enrich account, Boost Account, LinkedIn etc. But just imagine if they want to split Amazon for Kindle and Amazon for groceries separate (for example), wouldn’t it be annoying for us to have a different login? How are they going to understand your customer journey better and give recommendations that you know (just won’t admit) it’s been useful for you? Amazon’s user experience is top notch, makes our lives easier. My husband and I were loyal customers of Amazon when we were living in Boston. If only they can bring it over to Malaysia.

I think this leads to a bigger and more important point and work/research that’s currently being done which is on “user’s control of own data”, as highlighted by Fred Wilson on his twitter and blog. A proposed solution of a problem/concern that I am seeing myself that prevents companies to leverage on customers data for better use, for example to create a product/service that is tailored to your needs. It’s a privacy issue and with the tighter GDPR and more data breaches, it heightens the concerns and wipes out public’s trust. And for the more traditional companies, they would rather stay away than trying to work around it or problem solve this issue. How are they going to innovate?

I know MIT is very serious about this. They are trying to promote “better data security and privacy, while also allowing for easier data sharing, and more robust digital identity”. You can read all the whitepapers here if you are interested. It’s publicly available. There’s a silicon valley start-up called Helios Data who is collaborating with MIT on this.

The idea is to make users have full control of their own data. So we get to decide what data can be shared, when and to whom. It is after all our own data so we should be accountable for it.

Would you want to have full control of your own data? Would you want companies asking your permission if they can use your spending patterns details as part of their new product development? Would you? If you ask me, I would, but the process has to be simple and easy enough for me.

A “privacy-focused” Facebook or simply trust

If you are a Facebook user, which I’m sure you are, you should read Facebook’s or rather Mark Zuckerberg’s new business model as described in detail on his FB account. In a few words, Facebook will change from being a public social network to a private social network. This is what he said:

I believe a privacy-focused communications platform will become even more important than today’s open platforms. Privacy gives people the freedom to be themselves and connect more naturally, which is why we build social networks.


I believe the future of communication will increasingly shift to private, encrypted services where people can be confident what they say to each other stays secure and their messages and content won’t stick around forever.

One thing I can tell you for sure, the word “privacy-focused” is going to be used repeatedly by many people across the mainstream media (as it already is) and all social media platforms. I was curious to see how the potential change is being received. So far, there’s been more positive statements than negative as elucidated in the chart below (source: Brand24).


I’m still trying to make sense of his new plan especially on the end-to-end encryption. But one of the key principles highlighted that struck me was on Safety where encryption includes the privacy of people doing bad things. Which means, it’s even more difficult to catch the bad guys. In other words, it’s easier for the bad guys to do bad things.

There’s definitely a lot of work to be done before they can launch the new business model and there should be a balance between protecting for safety vs. protecting for business. If you put too much limitations/rules, the business environment will become too rigid for you to prosper. At the end of the day, it all comes down to trust. If only everyone’s trust can be measured.


A Good Board Member

Yesterday Fred Wilson wrote a post on a topic that I’ve been following closely. It’s about how to be a good board member. He linked it to another blogger, Mark Suster (who is also a VC, managing partner at Upfront Ventures) who wrote a list of what needs to be done before, during and after board meeting. You can read it here. It’s very useful.

I’ve mentioned a few times in my blog that my end goal is to be a good sounding board member of a company that requires my influence in the subject matter that I deeply care about. As what Fred Wilson said, his only rule of being a good board member is:


“If you care, really care, deeply care, like the way a parent cares for a child, you will be a good board member”.

At this point in time, I care about companies prioritising innovation and technology-related initiatives, especially the incumbents. I hear a lot of the talk, but most companies take a slow stance in walking the talk, especially at the management level and board. That’s what I saw when I worked on data analytics project with a few companies last year. Because it is something that you can’t see the fruit immediately, you have to grapple on the typical issues such as cost, budget, other priorities etc.

Anyway, two of the rules that Mark Suster highlighted are consistent with what I’ve read about being a good board member so far. The ineffective ones or rather those that couldn’t make the other board members agree to them or influence them are those that do not practice the rules. They are as follows:

  1. Speak with the CEO before the board meeting
  2. Have calls or emails with other board members before the board meeting

I could relate to this because it’s not just applicable for board meetings. but also the day-to-day job in the corporate world. When you want to make changes, you have to engage and syndicate with the relevant people first. Get their buy in so that you can have a few people supporting you in a larger meeting/discussion. When I was in MIT, the professor taught us to find the ‘nancy’, i.e. the one who’s mostly connected to the people that you need their support. I’ve seen a few people that just don’t believe in this. They think it’s a waste of time. I hope these people realize one day how important to engage.