A national strategy for artificial intelligence?

Tuesday, 9 April 2019 00:00 -     - {{hitsCtrl.values.hits}}

Colombo is talking about artificial intelligence. Millennium IT Founder and other companies Tony Weeresinghe delivered an oration on AI to the alumni of the University of Colombo. Vallibel One Group Founder Chairman and power behind many others Dhammika Perera delivered a two-hour monologue on the subject at an event organised by the Computer Society of Sri Lanka a few weeks back.  

These are good. To become a knowledge-based economy we have to understand and ride the waves of technology that keep rolling into this island. Beyond that, we must contribute to the world’s knowledge.

Telegraphy was introduced to this country in 1858, within 20 years of commercial deployment in the UK.  The first license for mobile telephony was given in 1989, well before our neighbours got started. An exception was the bipartisan delay on introducing TV until 1978.

With network technologies made up of multiple interdependent elements, there was a choice. If the state does not make the initial investment decisions or assign the frequencies, the network technology stay outside.

Why a national strategy?

But things like artificial intelligence (AI) and connected devices (also known as Internet of Things or IoT) are different. They are embedded in other things like map programs or motor vehicles, which do not require licenses and frequencies and are therefore difficult to exclude. I recall survey findings from some years back from Thailand. The frequency auction had been delayed but people were already using 3G enabled phones, which of course did not permit optimal use.  

There is nothing to stop a Sri Lankan with hearing disability from using Google’s Live Transcribe (recently introduced in over 70 languages). That means there is nothing to stop data from that person’s conversations from being used to ‘train’ the AI that powers the app that helps a disabled person understand what is being said. The point is that AI is creeping into our lives whether a national strategy is in place or not.

So why does the country (or a company) need an AI strategy? It needs a strategy either to proactively adopt AI in its business processes or to position itself as a supplier of AI technology to other users, or both. The two events that I participated in last month emphasised the former. Surprisingly, these leading businessmen had little to say about the supply side, about positioning the country as a centre of excellence in the supply of AI.  

We first talked about adopting and using computers back in the 1980s; it is later that we addressed the supply side. Now we have gone much further and have national strategies on how to increase export earnings from IT and IT enabled services. We have to address the supply side as well, perhaps with less of a lag. But these people got the conversation started; let us not quibble.

Demand side

What is AI? It is machines that show some behaviours that mimic human intelligence. These days, what we have are narrow AI in specific domains. It is based on deep learning wherein the software is trained on massive amounts of domain-specific data. 

Examples of what AI can do are make decisions/or advise those who are making decisions on creditworthiness; assist judges on bail and sentencing, based on correlations that suggest likelihood of recidivism; or make diagnosis of medical conditions more efficient. The possibilities are endless.

The rules by which the software reaches its conclusions are not understandable to humans, so the results have to be verified against ground truth. Issues of bias or error have to be addressed. What works in California or with white males of a certain age group may not necessarily work in Sri Lanka or with brown-skinned women. AI engineers are needed to customise the software for specific conditions.  

So it’s qualitatively different. One could just buy spreadsheet software and use it with a little training.  But AI requires skilled engineers to customise for best results (and at least to minimise errors). In some cases, additional training data may be needed. Where are these engineers? How will they learn about AI? Where are the data? Are they in data form? When the data are about people, one would also need safeguards for the data.  

All these elements could be included in a well-formulated national strategy. But sadly, these elements were not addressed in either of the presentations. This is perhaps why the Computer Society should have kept to the original plan for a conference with multiple speakers and exchanges of views. 

Data analytics to AI

Before AI, big data and data analytics were the buzzwords. It was possible to ask a consultant to, for example, run a company’s customer records through a ‘black box’ proprietary software to identify the most valuable customers, predict the ones most likely to defect and so on (today, these things are likely to be packaged as AI). Data cleaning would most likely be necessary.  

Because the whole thing would be done in-house, data protection issues were unlikely to crop up. It would be good to have a data scientist on staff, but not essential. Though the underlying software would be open source, the consultants would have little incentive to share the secret sauce unless the company or the in-house data scientist was insistent. So the costs would be somewhat high.

Things would get more complicated when the company starts working with external data sets, such as when it seeks to gain insights for marketing. Here, issues of representivity (does the data accurately depict the target population?) and also limitations if any on how the data may be used (for example, is the data pseudonymised or anonymised? If the former, patterns can be identified, even if the individual cannot be) arise. There is now a greater necessity of domain knowledge and possibly also for in-house expertise in analytics.

For AI, training data is critically important. Many experts believe that China is likely to take the lead in AI because of the greater availability of training data (China’s digitalisation is highly advanced, for example in payments and facial recognition). Europe is likely to lag behind because of the excessive restrictions on data use and consent requirements.

Supply side

Barriers to entry are low in most IT domains. People did not know what open source and middleware were two decades back, but Sri Lankan entrepreneurs were able to quickly carve out niches in these areas. Repeating these successes in AI is not outside the realm of possibility.

Given the need for training data and skilled AI engineers even on the demand side, it makes little sense to simply focus on deploying AI within companies without also exploring the opportunities of becoming suppliers of AI solutions and AI-infused products. 

Now, when adoption of data protection legislation appears imminent, the time is opportune to adopt a national AI strategy. Otherwise, we may find AI foreclosed by short-term considerations associated with doing BPM business with Europe.