Artificial Intelligence and Data Analytics
The global race to augment capabilities of artificial intelligence (AI) is intensifying in both advanced and emerging economies. From optimising power generation and transmission, diagnosis and drug discovery, improving learning environment, enhancing design and functionality, to automation in logistics, AI will not only continue to evolve but possibly surpass human intelligence in the near future. Today’s digital age is overflowing with valuable data which if appropriately analyzed can predict results, making data analytics an indispensable tool for any corporate to sustain in the economy.
AI tools encompass deep learning skills that train machines to derive output from data fed into it without any human involvement, mirroring a human’s ability to learn from experiences. Advanced analytics has emerged as an influential tool to learn from past performance to more accurately predict future trends. The impact can be significant when asset owners are able to predict the likelihood of a given event, by resorting to drones, robotic systems and Internet of Things (IoT) for information, and using it to update the other processes, take preventative mitigation measures, reduce scheduled down-time, know customer preferences, and focus on maintenance rather than replacement.
In the specific context of infrastructure investments, utilisation of effective data analytics tools can assist infrastructure companies in accurate risk profiling and redesigning of portfolios leading to smarter investment decisions. This in turn will push economic growth with healthy competition through optimised processes, augmented employee efficiency and enhanced customer experience.
Interface of Data Analytics, AI and Infrastructure Investment
For an infrastructure company, the key step prior to making an investment in a specific project is to assess and evaluate all the internal and external data available on design, construction, operations, maintenance, capital expenditure, incidents, historical failures, competition, inspection etc. Predictive analysis of all such data available is crucial to bidders, investors, developers and lenders, for understanding the viability of project investment, studying consumer behavior and anticipating future change. Post commissioning, an asset owner should focus on collection of data to utilise the same in proportional hazard modeling, predictive maintenance, incipient failure detection, probability distribution functions and IoT, which plays a big role in identifying structural defects, failures, life cycle and required maintenance of the assets.
Decisions of the asset owners, in both the public and private sectors, hinge on the trade-off between continual investment in maintaining an asset compared with replacing it, and presently many such decisions are being made without access to the kind of analytics that could be best suited to make such decisions. Application of advanced analytics approaches will assist in generation of deeper insights and value on infrastructure life cycles, maintenance versus replacement decisions, and asset longevity trends. Decisions based on data driven methods will eventually assist the organisations to recognise image, improve decision making, deploy capital appropriately and free up capital as and when required, thus allowing for its reallocation to other attractive opportunities.
AI models such as blockchain and cryptocurrency have proven to be significantly helpful in reducing infrastructure investment cost, tracking transactional data in digital ledgers, executing smart contracts in a decentralised data environment, reaching environmental sustainability and making efficient decisions. For instance, blockchain model has become instrumental in the 5G infrastructure investments, because of its features of decentralisation and transparency, with potential reduction in costs. Recently, the Brazilian government has incorporated the blockchain and cryptocurrency technology in Sao Paulo to pay engineers, involved in conducting feasibility studies to assess the viability of infrastructure projects, with buildcoins. These buildcoins are redeemable for other services in the same ecosystem, such as subcontracting, market research and professional training. Such innovation seeks to bring the infrastructure community closer by, attracting experts from across the globe and making the process more transparent, faster and cheaper.
Key Sectors Adopting Data Analytics
Traditionally, quality of infrastructure was measured by its lifespan, which determined the well worth of the investment made in infrastructure. But now, infrastructure has become intelligent, is able to communicate with other systems and probe various infrastructural sectors, via AI tools, in ways that were unknown a few years ago. We have seen some interesting uses across various sectors:
- Telecom – With the introduction of text analysing chatbot TOBi, Vodafone witnessed a 68 percent improvement in consumer satisfaction. Similarly, Nokia witnessed 20 percent to 40 percent improvement by using the virtual assistant MIKA, who suggests solutions to network issues. Nokia also launched its own machine learning-based cloud AVA platform, to better manage capacity planning and predict service degradations.
- Railways – Data analytics help in determining routes, service price, freight management and safety measures. AI-enabled robot called ‘USTAAD’, developed by the Indian Railways (IR), scrutinises under-gear parts of the train and monitors the railway coaches and engines to carry out repair in cases of faults, thereby enhancing safety. AI is also being deployed to assist Indian Railways Catering and Tourism Corporation (IRCTC) in ticketing management, e-catering, optimising operations, route and schedule planning, and resource allocation.
- Airports – Bengaluru International Airport Limited has signed an agreement with Unisys Corporation to develop an analytics center for excellence, which will provide business intelligence and advanced data analytics platform, to assist the airport staff to enhance the experience of travelers by providing real-time flight information, communication of airport services and personalised retail offerings based on passenger preferences and past spending history.
- Real Estate – IoT sensors are used to monitor buildings, dams and bridges, to identify outages and breakdowns. AT&T launched Smart Cities framework to help cities like Atlanta, Chicago and Dallas by using IoT innovations, bringing transformation to public safety, transportation, infrastructure, energy conservation and forming alliances with technology leaders like Cisco, Deloitte, Ericsson, GE, IBM, Intel, and Qualcomm Technologies, Inc. to build and support more connected framework.
- Roadways – Netradyne has developed an AI to understand the distance between fellow drivers, traffic rule monitoring and law enforcement. Siemens Mobility has built a prototype monitoring system to estimate the density of the traffic and alter the traffic lights by such real time congestion status, identified by using the traffic cameras and informing the same to the back-end control center. For managing the traffic in restricted zones, AI allows parking by facial recognition, sending an alarm in case of a face not matching with the database.
- Ports – Adani Ports and Special Economic Zone has set up AI for threshold setting and routine centralized monitoring in a cloud-based system, helping them to focus on business issues rather than maintenance. High quality information technologies, being able to predict breakdowns beforehand, provides them with a major competitive edge.
Limitations of Data Analytics and AI tools
Whilst it is beneficial to use AI tools, over depending on the same has its own limitations and risks. Machine learning model not only reflects human cognitive bias but also reinforces its own learning based on the bias and sometimes generates results that are incorrect and undesirable. The efficiency and effectiveness of an AI tool is entirely dependent on the completeness, authenticity and integrity of data. Therefore, it is imperative that the underlying data needs to be inclusive and complete and the AI tools needs to be defect free, in order to avoid delivery of biased output. Above all, are the concerns that with AI in place, humans will not be required to manage and monitor machines including regulating input and output, resulting in job losses.
Additionally, there are always concerns about the right to possess data vis-à-vis the duty of maintaining its privacy. Data selling has become a huge market not only in India but across the world, with both the private players as well as government bodies involved. Under the Smart Cities Mission, the India Smart Grid Forum (ISGF) and the India Smart Grid Task Force (ISGTF) were set up to frame policies and roadmap for the installation of smart metres, which collects, measures, analyses and monitors real time energy usage, in short intervals and communicate the same to the service providers. Smart meters also communicate with other appliances and enable consumers to understand their consumption pattern. However, the problem is that these meters also have the ability to track the exact pattern of activities performed by an individual, thus posing a threat to a possible privacy invasion. Sourcing and use of data without any violations, especially in the absence of a comprehensive regulatory framework itself has become a big challenge. While there is enough data available, the place of origin, its use and dissemination of the processed information can always pose a risk for data analytics from a compliance perspective.
The Information Technology Act, 2000 (IT Act) and the rules framed thereunder seek to protect personal information by regulating its collection, receipt, possession, storage or any other manner of dealing or handling. In the case of Justice K.S. Puttaswamy (Retd.) and Anr. v. Union of India, (2019) 1 SCC 1, the Supreme Court recognised the right to privacy including informational privacy as a fundamental right enforceable against both the state and private parties, and emphasised on the need of comprehensive data protection laws in the country. As a result, Personal Data Protection Bill, 2019 (PDP Bill), significantly influenced by the European Union General Data Protection Regulation, was recently presented in the lower house of Parliament, seeking to establish fiduciary relationship between the processor of personal information and the individual whose data is being processed.
The duties and obligations between the licensor and/or the author of the AI and the end-user of the AI tools typically flow from a contractual framework, depending on the nature of the transaction. Contracting between the developer and the user organisation for using any AI tool, typically relies on the negotiations, bargaining power, uniqueness of the product, competitors in the market and buyers possessing knowledge of the product. Various considerations for negotiating such contracts and avoiding conflicts in the future depend on the parties deciding beforehand, the underlying product of the transaction (transfer of technology versus right to use), licencing (exclusivity and transferability), usage of the product (internal or external), payment mechanism (consideration only for the sole product or inclusive of machine learning add-ons), maintenance and support services, term and termination (perpetual transfer versus temporary transfer), confidentiality (both of the product and the data of the end-user entered into the product), warranties provided by the seller (for defect free product considering the AI tools learn from data entered by other users), intellectual property rights and dispute resolution mechanism (considering legal framework in India is still evolving).
Aspects to Watch Out For
- The legal personality of AI is still being debated globally.
- The legal framework governing AI is still evolving in India. Allocation of liability in case of accidents is unclear – would it lie with the government, the investor company or the AI model, considering that it functions on its own machine learning.
- Neither the IT Act nor the PDP Bill makes expressly and conclusively clear the question of ownership of big data. The attribution of liability for machine learning and deep learning is still unclear. The IT Act provides for no liability for government agencies such as ISGF and ISGTF which possess large volumes of data in relation to the infrastructure projects.
- For technology developers, it is important to note that while AI algorithms are protectable as literary works under the Indian Copyright Act, 1957, this protection is not adequate because the functional aspects of the algorithm or any subsequent variations made to it are not protectable.
- AI tools are like ‘black box’ and the reasoning behind a particular outcome of an AI tool is usually unknown to the users, thus making the regulatory and judicial dialogue with the developers and other stakeholders very crucial to deliver an ethical outcome.
- The sectoral regulators’ approach towards AI, considering the increase in interface between the two, is something to look forward to.