Submitted by Kshitij Naik and Ankur Pandey, Associate Editors, The Indian Learning & Vasudha Tiwari, Research Intern.
In this era of rapid technological advancement, Artificial Intelligence is poised to disrupt the businesses and industries before long. While it was perceived that AI advancements will primarily affect the Tech Industry, a survey of 1004 Senior Executives (across sectors and geographies) suggests that organizations in other sectors are not hesitant to test the waters. Among the respondent organizations, 87% had begun deploying AI by 2019, and by the end of 2020, the numbers will go up to 97%. The IT and Telecom firms have been precursors with 81% of them using AI by 2018, followed by financial services firms (78%) and consumer goods and retail (75%). AI is being used in various operations such as customer service, fraud detection, quality control in various industries and Government Organisations. While it is certain that AI will have a different role to play in different sectors, we need to resolve the issues associated with scalability, data sharing, data privacy, and AI bias to be able to reap the maximum benefits of technology.
AI IN BUSINESS
The survey provides insights into the role, impact, and future of AI in various businesses. Some of the highlights of the survey are:
Return of Investment: Across sectors, the respondents agree that the Return on Investment of using AI has been either at par with their expectations (59%) or exceeding their expectations (37%). While early adopters were more likely to have the ROI exceeding their expectations (54% in 2015 compared with 13% in 2019), they were also more likely to have been disappointed with it (10% in 2015 compared with 5% in 2019). This reflects that while the first movers were likely to get more returns than expected, the experience has reduced the chances of failure for new entrants.
Accomplishments: The returns from the use of AI have been in the form of operational efficiency and increased savings (cited by 51% respondents), better management decision-making (44%) and improved customer experience (41%). The top AI use cases across sectors have been quality control, customer care, and fraud detection, though there is considerable variety in use cases pursued by different sectors.
Iterative approach: Since the actual and assumed capabilities of AI are different and the challenges may be in the form of lack of technology limitations, high costs, lack of data, early attempts may fail to make it beyond the pilot stage. But these challenges might be resolved later, making the project viable. Cancer therapy using AI is one such example that was not feasible a few years back, but with the latest technology and relevant data, it has become scalable.
Future opportunities: The growth of Edge computing - in which the computing takes place at the edge- near the physical location of the user or source of data, will reduce the need for expensive cloud servers. This will make real-time video analysis techniques scalable and usher a revolution in personal AI, with even digital replicas of humans possible. This will transform the health, education, and entertainment sectors with the ability to gather and analyze real-time data of customers’ behaviour using which the businesses can tailor products suited to each individual.
CHALLENGES AHEAD
Technological & Skill Deficiency in AI: Existing technology limitations and regulations in Industries can hinder the wider adoption of AI to a large extent. For example, advances need to be made in natural language processing before chatbots become more sophisticated which they are not at the moment. Another major constraint on the use of AI is the shortage of internal data scientists and related experts, which is felt keenly in the manufacturing and technology industry. Companies building advanced AI models internally are often constrained by a talent deficit.
Insufficient understanding of integrating data: Many organizations want to integrate AI into their system independently but they fail to understand the value of integrating data into a single platform from where they could train their independent AI Models. Since there exists very little regulation in relation to Data Processing and storing, not many organizations would like to put their reputation at stake and share their data or integrate it with a central system.
Unstructured data and sourcing data from open-source platforms: One of the major challenges before businesses who want to introduce AI into their systems is unstructured data. In order to train AI systems and for the systems to work efficiently, the data needs to be in a structured format. Even though organizations have a large amount of data, their ability to extract value from this data is really limited at this point in time. Organizations relying on a variety of open-source platforms is also a major cause of concern due to the unreliability of such data.
AI Ethics: A number of technology companies have received heavy criticism from policymakers, non-governmental organizations, and the media over their lack of concern for the ethical use of AI. Even bigger Tech giants like Amazon have often come under the radar for developing AI tech unethically, hence it is suggested that every company ought to have an Ethics Board.
THE SILVER LINING: CASE STUDIES
Walmart's AI-powered store: Walmart has set up an Intelligent Retail Lab (IRL) at its store in Levittown, New York in 2019. This store acts as an R&D lab for Walmart but it is set up in an actual working store. The AI-powered system uses a large amount of data from the intelligent cameras and various sensors in the store to intelligently determine when certain shelves need restocking, when fresh foods need to be replaced and help to track the inventory. The store not only helps them in their R&D but also helps them educate their Customers about AI systems. However, what we need to understand is that a similar model cannot be implemented in many other places due to legal constraints and also due to bandwidth and computing power limitations in most of the stores. Thus, even before we develop the computing power required to adopt many more such models, we need a much more sophisticated legal and ethical structure in place.
Pharmaceutical industry’s initiative for data sharing: A few large pharmaceutical companies have formed a consortium to share data and train their AI systems using Federated Learning to train their algorithm. This model does not require them to exchange their data sample, thus maintaining the anonymity of the data. Such sharing of data has led to simplified and accelerated drug discovery and development, resulting in new and less-costly drugs and treatments reaching the market.
WAY FORWARD
It is certain that AI technologies are here to disrupt the businesses and there is a lot to be gained from it. But we need to build the capacity to resolve the issues that come along with it so that we can maximize potential benefits from AI-based tech solutions. Data sharing on a scale that leads to new AI-enabled efficiencies, products, and value chains is a vision yet to be realized. It will surely come to pass, but that it will take time before apprehensions about the security and privacy risks of doing so are eased. The laws and regulations need to change or be clarified and simultaneously industry standards need to be developed before the firms can embrace data sharing widely. Data trusts or banks can be established to facilitate data sharing. Machine Learning Techniques like Federated Learning that was used by the Pharmaceutical Companies could be leveraged too. AI can be trained better when there is a large amount of data from varied sources, instead of integrating data from open sources as such data at times may be biased or may have various issues with them. Also, governments around the world need to set certain ethical and legal guidelines surrounding AI as soon as possible in order for organizations to have more confidence in building AI-based systems. Organizations need to realize that they need to put data privacy of their customers as their top priority. The future of AI in businesses is indeed promising, but we must be wary of our steps to maximize the benefits and minimize the potential harm that AI technologies can bring about.
To understand the development, refer to https://www.technologyreview.com/2020/03/26/950287/the-global-ai-agenda-promise-reality-and-a-future-of-data-sharing/
Mail us for queries at editorial@isail.in.