Interview with Fusionex’s Raju Chellam
EITN: The recent privacy scare with Cambridge Analytica, jolted a lot of people into realising how much information they actually share in exchange for service and convenience. Do you see people being more careful about sharing their data now? Will this impact big data analytics (BDA) and artificial intelligence (AI) solutions?
Raju Chellam: The fallout from the Cambridge Analytica (CA) scandal has become a watershed moment, not just for social media, but also for the corporate world, and for the key companies involved.
The Guardian newspaper headlined its story as “Facebook’s week of shame: the CA fallout” and noted that “US$$60 billion was wiped off Facebook’s market capitalisation in wake of Zuckerberg’s silence over data breach.”
CA’s parent, SCL Elections Ltd, announced in early May that it was filing for insolvency in the UK and the US and closing all of its operations. The London-based company said CA had faced “numerous unfounded accusations” and been “vilified for activities that are not only legal, but also widely accepted as a standard component of online advertising in both the political and commercial arenas.” However, corporate registration documents in London indicate that several SCL and CA executives have launched a new venture called Emerdata Ltd, The Globe & Mail reported on May 2, 2018.
Has this been a wakeup call for social media users? Will it impact BDA and AI solutions? Yes and no. Yes, for literate and informed users, mainly in countries directly hit by data captured/analysed by CA. No, for the millions of users across Asia, Africa, Latin America. As well as a few other millions who don’t quite care. After all, it’s not just social media that’s capturing your data.
Take loyalty cards for example. Most consumers know that loyalty cards are used to track their behaviour and that the data is sold to marketers. Would they stop using these cards if they knew? “Research shows that people in surveys say they want to maintain their privacy rights, but when asked how much they’re willing to give up in user experience – or to pay for it – the result is not too much,” an article in the Knowledge@Wharton noted. In other words, there’s a difference between how we care about privacy as an idea, and how much we’re willing to give up to maintain it.”
EITN: The fact that consumers share their data is what makes BDA and AI for enhanced service and convenience, possible. How can technology and/or process help meet the need to collect more customer data, and also the need to protect customer privacy?
Raju: BDA and AI are not yet at an inflection point where it is possible to either collect all the data, or to use all of that data for what is called actionable intelligence. On the other hand, not all data captured either directly or indirectly is harmful. However, after the Cambridge Analytica scandal broke, customer privacy became a hot potato.
In April, two US Senators (Democrats) introduced a privacy “Bill of Rights” to protect US consumers’ personal data. Called CONSENT (Customer Online Notification for Stopping Edge-provider Network Transgressions) Act would require the FTC (Federal Trade Commission) to establish privacy protection for customers of “edge” providers – or entities providing direct connection/apps with the user – like Facebook and Google.
Like the GDPR (General Data Protection Regulation) in the EU, the US CONSENT Act wants companies to obtain opt-in consent from users to use, share or sell users’ personal information, and notify users about all collection, use and sharing of users’ personal information, as well as notify them in the event of a breach. These practices are being made into regulations already in countries like Malaysia and Singapore, and will soon be adopted worldwide.
On the flip side, not all data captured is harmful. For example if there’s a major fire in a specific locality in KL where many people can be potentially trapped or burnt, LBS (location based services) and identification of people to alert them can save thousands of lives. The same goes for infectious disease prediction and management, and for tracking criminal activities.
EITN: Can AI and machine learning actually enhance privacy, and do you have any examples of this you can share?
Raju: The right question should be: Can AI and ML (machine learning) diagnose, monitor and/or prevent the abuse of privacy and security? It absolutely can. This can be done both by BDA which can throw up the analytics for the AI engine to test the accuracy for “false positives”. As stated earlier, not all analytics is detrimental. So the key is not to enhance privacy for its own sake, but to ensure that the “actionable intelligence” is not being abused – nor any legal regulations breached – for commercial gain.
Another key question: Can AI save lives? AI engineers at the Houston Methodist Research Institute developed software in 2016 that can accurately diagnose a patient’s breast cancer 30 times faster than doctors could. When fed mammogram results and medical histories of about 500 patients, the software diagnosed breast cancer with 99% accuracy, according to an IEEE report. The software also produces fewer false positives than doctors do. “More than 20 million people each year are sold worldwide into prostitution,” according to a Fortune article. “Researchers at the University of California, Berkeley, have developed AI tools to identify sex-trafficking rings, making the leaders easier to target and prosecute.”
EITN: What are the top 3 uses of AI in Fusionex’s portfolio of offerings and services?
Raju: Fusionex has developed an AI-powered chatbot which has been rolled out for a telecommunications service provider. This technology solution leverages on the public’s rising preference to communicate more on mobile via messaging apps. Being a virtual assistant, the chatbot does not require rest or experience any downtime. The system is available to interact 24/7 and is made available for access via a Web Chat function supported by any Internet browser and also through a mobile app.
The AI deciphers user questions using the system’s Natural Language Processing feature and provides answers to user queries quickly, efficiently, and accurately. To provide a more personalized service, the chatbot was augmented with data analytics which allowed the virtual assistant to mine data regarding user habits. The chatbot could then recommend products or services to users based on their preferences. This increases the chance of the promotions being taken up and improves customer satisfaction knowing their needs are being tended to attentively, which bolsters brand loyalty in the long term.
Fusionex has also developed AI solutions for the manufacturing sector, which has been rolled out specifically for an electronics manufacturer. Designed to bring new business intelligence to its manufacturing capabilities, the solution connected and synchronized data sources for real-time monitoring as well as predictive and prescriptive analytics. By predicting potential failures of equipment, the solution could intelligently notify and prompt replacement of parts, before they fail, increasing manufacturing throughput, increasing yields, improving efficiency, and reducing downtime. The AI solution could also identify defective units 10x faster than manual inspections, saving time, and ensuring product quality.
AI-powered image recognition technology has also been deployed in cross-border trade initiatives. Customs departments use this solution to alert their officials should prohibited items be identified.
EITN: How do your customers prepare to use your solutions? How do they prepare their data to be ingested/processed by Fusionex?
Raju: We offer a wide range of connectors available for our clients to connect our solutions to their system. This is so that our customers can feed their information easily and directly into the system as well as reducing manual data keying. This makes it possible to bring together an integrated and unified system, which includes our customers’ legacy systems. Fusionex enables customers to consolidate data from disparate sources into one native platform for easy cleaning, processing, and analysis. These vast variety of sources Fusionex may connect to include local files, databases, cloud services, third party applications, and more. Some of the formats include spreadsheets, CSV, Flat Files, SQL Server, PostgreSQL, MySQL, DB2, Oracle, Hive, Spark, and SQL Azure.
EITN: What is next for AI in Fusionex’s roadmap, for the next 3 years?
Raju: The AI roadmap for Fusionex is both exciting and expanding. That’s because the market for AI itself is skyrocketing. In 2017, the global market for AI products and services was worth US$12 billion, up 59% over 2016. By 2021, the market will cross US$57.6 billion, according to estimates from IDC (International Data Corp). That’s a CAGR (compound annual growth rate) of 50% between 2016 and 2021.
Moreover, Fusionex leads in BDA solutions for the retail and banking segments, which coincidentally are the two biggest and fastest-growing markets in AI. Who are the likely big spenders on AI and BI? Retail and banking top the list, followed by manufacturing and healthcare. Their combined investments will account for up to 55% of all global spend on AI, IDC says.
“The retail industry invested about US$1.74 billion on AI last year,” IDC notes. “The BFSI (banking and financial services industries) likely invested another US$1.72 billion on cognitive systems and software. Between now and 2021, the biggest spenders will be the retail players.”
One key challenge? Talent. The McKinsey Global Institute quantified in 2011 that the US alone will be short of 140,000 to 190,000 data scientists – or people with “deep analytical skills” – by 2018. Now McKinsey forecasts that millions of people will be needed to serve as translators of the results of the work of data scientists to the rest of the organisation.
“It may be easier for domain experts, with deep knowledge of the business in which they are engaged, and the requisite interpersonal skills, to obtain sufficient knowledge about data analysis to act as the translator for data scientists,” Forbes magazine quoted The Sloan Management Review’s report. “That’s as compared to data scientists trying to gain enough knowledge about the domain, especially the language of that domain.” It would thus be easier for healthcare professionals, for example, to learn about data analytics, than for data engineers or software developers to grasp the complex and intricate world of medicine or genetics.
How does Fusionex deal with the talent shortage? By doing its own training under the Fusionex Academy which offers practical, industry focused, and HRDF claimable professional training courses and certifications ranging from Data Analytics, Data Engineering, Data Science to Python Programming. Other than its own academy, Fusionex also collaborates with partners such as PIKOM and universities including but not limited to Asia Pacific University of Technology & Innovation (APU), University Malaysia of Computer Science & Engineering (UNIMY), Universiti Technologi Malaysia (UTM), University College of Technology Sarawak (UCTS) and Universiti Malaysia Sarawak (UNIMAS) to train working professionals and students to be data professionals, in line with the Government’s goal of achieving 20,000 data professionals by 2020.