It's anticipated that we are soon to live through the fourth technological revolution. A recent report by McKinsey revealed that AI is predicted to be utilized in at least one area of business by 70% of organizations by 2030. The technology is a favorite topic of conversation for people from all walks of life, as it promises to entirely revolutionize our lives and business. As Baidu Chief Andrew Ng famously stated in 2017: "AI is the new electricity."
But will AI’s effect really be so significantly more impactful than all other technologies that came before it?
To find out, ahead of Innovation Enterprise's AI & Predictive Analytics Summit we sat down with Partha Dutta, SVP of data science at Sembcorp to talk about the impact AI will have:
Innovation Enterprise: What are some of the biggest dangers of using AI to make business decisions?
Partha Dutta: The purpose of AI, in a broad sense, is to help decision making. The circumstances under which we make choices vary - from the simple and mundane (e.g. selecting our favourite song from a playlist) to highly complex and higher-risk (e.g. the real-time control of a mission-critical system). For those applications of AI that tend towards the latter category, adequate analysis of the risks, costs, and benefits must be conducted. Complex systems typically have a large number of "states", and for an automation (i.e. AI) to function reliably in all of the system states, it should be adequately field-tested before deployment.
Perhaps since AI is still a relatively novel area of science, we tend to think about it differently. But isn't it the case that for any new technology or product introduction, a business does a cost-benefit due diligence analysis? Understandably, for AI, this might be a new challenge for most, but it can be managed with the right vision and skill.
IE: What technology has made the biggest change to your work in the last two years? How has it changed the way you work today?
PD: I would have liked to answer this question with an example of one of the more popular topics in AI or Machine Learning (ML)! While the innovation and embedding of new AI and ML technologies is ongoing, it takes time to fully realize their benefits at scale, especially in complex business environments.
What has really changed my work in the short term, however, is the use of advanced collaboration tools which effectively connect distributed teams and help them overcome the barriers of time-zone difference, distance, and other social factors such as lack of face time.
As an example of one such collaboration tool, we use Slack to connect the project delivery and business teams. It helps to resolve issues and enable more efficient decision-making.
IE: What trends do you predict are most going to affect predictive analytics in 2019?
PD: In my own experience, there is sufficient richness in the fundamental algorithms and software as a service (SAAS) offerings to meet the requirements of most business applications. One of the gaps is in the ability to effectively transform these capabilities from model development to deployment of analytics products that can generate measurable business benefits at scale. This can be achieved by not just investing in the right technology but also in the right organizational structures and processes.
This is also not confined to data science alone, but has an impact on related technological disciplines, therefore creating a rich ecosystem of co-existence. As an example of the productization of AI solutions, Sembcorp’s data science team works very closely with our group IT department and global business lines to develop a technology roadmap from initial model development through to deployment and ensures that the business owns the productizing effort. This approach has been one of the reasons for our success in deploying a number of solutions within a short period of time. These include a cloud-hosted data lake as a single-source-of-truth for our global operational data amounting to about 200,000 sensor feeds in near real-time, an automated root-cause analysis solution to optimize plant performance, and an effluent quality prediction solution for industrial waste water treatment.
Different industries are addressing this need in different ways - depending on their business strategy and stage of adoption of AI - and this will continue to be an important area of focus for businesses committed to achieve the full potential of AI.
Partha Dutta, SVP of Data Science at Sembcorp is speaking at Innovation Enterprise's AI & Predictive Analytics Innovation Summit in Singapore on October 10–11, 2018