Understanding Artificial Intelligence (AI) for business application of Narrow AI (ANI)
AI is a prevalent topic of discussion, specifically focusing on what AI tools like ChatGPT and Bard will mean for people's careers and how they will transform their ways of working. As a business strategist deeply passionate about innovation, I aim to delve into how the adoption of AI will revolutionise our service approaches, particularly focusing on Narrow AI (ANI), the most mature form, and expanding into some General AI (AGI) hypotheses. In this post, I will examine its application across the banking, travel, and retail supermarket industries.
As this marks my first post on Artificial Intelligence (AI), I will begin by offering an introduction and establishing context for not only this discussion but also for future posts as I delve deeper into this domain.
In my role as a strategist and innovation professional deeply passionate about Human-Centered Design (HCD), I have leaned in to comprehend how AI will shape changes in our lives and alter the way we interact with organisations to access and engage with their products and services. This exploration has led me to contemplate its applications for organisations and how we are likely to seamlessly integrate AI into our daily lives, driven by the rapid adoption of innovative technologies that enhance convenience and efficiency. In fact, this transformation is already underway we are already seeing its application, and therefore we will witness a growing imperative for organisations to formulate a well-defined AI strategy or find themselves challenging to compete.
So what is Artificial Inteligence (AI)
Today it is viewed as a simulation of human intelligence processed by machines, usually computer systems which can evolve through its ability to learn, reason, problem solve, use of perception and language understanding.
The 2 types of Artificial Intelligence (AI) - Narrow and General
Today there are two types of AI, and these are classified as Narrow (weak) or General (strong).
Narrow AI (ANI)[weak]
This refers to AI that is meticulously designed and trained for particular tasks or within confined domains. These AI systems demonstrate exceptional proficiency in executing a pre-defined set of functions and cannot transfer knowledge or skills to tasks beyond their designated domain or activities for which it performs. I want to emphasize that the term "Narrow AI" does not imply inferiority. For most organisations, the implementation of Narrow AI will unlock substantial opportunities, providing both tangible business and customer benefits, provided you have a robust strategy and the necessary data to support it.
The use of Narrow AI will be the focus for most organisations today, that will focus on the material gains and service evolutions to accommodate and realise the value in its introduction and maturity over time.
General AI (AGI) [strong]
This term refers to AI systems possessing a level of intelligence deemed comparable to, or even surpassing, our own. AGI demonstrates the ability to comprehend, learn, and apply knowledge across diverse domains, geographies, and systems, showcasing a broad spectrum of cognitive abilities. As per my current understanding, the application of AGI is still a considerable way off. However, for comprehensive coverage, I will supplement this blog with some hypotheses.
Our adoption of new technologies is moving at a pace that has never been experienced before, which is why organisations need to lean in to understand how AI could benefit them in driving buisness and customer value.
Some Working Examples - across Banking, Travel and Retail Supermarkets case studies.
These examples give some context based on my hypotheses of how things may be performed today without AI and what could be a possible future state with Narrow and General AI. Whilst for many, these will be exciting opportunities, they will require an AI strategy to see their successful implementation and a clear understanding of the readiness to apply them based on your organisations current state.
A disclaimer - These are provided as generalisations and don’t take into any specific organiation or their maturity or the industries maturity for which they operate in, and does not assess constraints that exist today like regulations or standards.
Banking examples
Fraud & Scam Detection
In banking today, the systems for fraud detection may be rule-based and trigger workflow or reporting processes.
Narrow AI -Could use machine learning to identify suspicious patterns. As it matures, it could provide advanced real-time detection based on its learning of fraud and scams within its domain of operations.
General AI - Could use advanced fraud detection algorithms to detect fraud and scams across multiple financial service providers, networks, geographical regions and financial markets and real-time detections considering complex patterns. Maturing by optimising its understanding of the types of activities, creating real-time alerts, tactical controls, and informing strategic controls or mitigates. It also may have the ability of traceability and be able to detect its source.
Customer Service - Chatbots & Live Chat
Today we have automated Chatbots or human-powered Live Chat supporting digital channels. Most Chatbots have limited capabilities, even for simple banking transactions, and will generally see customers passed to Live Chat or other channels when thresholds are hit. Live Chat is usually limited to the .com internet sites and has limited tolerances in their support, and very few have Live Chat on their internet banking platforms.
Narrow AI - Could be deployed and start learning to support scripted responses across all digital platforms within its domain. Later enhanced learning and Natural Language Processing (NLP) would allow for better conversation flow and contextual understanding, allowing customers to stay in the channel of choice and meet their needs.
General AI - By not being limited to a single domain, chatbots will have a broader contextual understanding. It will see a fast-paced transition to conversational AI with near-human understanding for seamless interactions and outcomes. Will have the ability to provide a global market understanding of financial markets, products and services and jurisdictions of operations. For example, someone wanting to know how to set up a transaction account in a foreign country could be supported in identifying a suitable institution and criterion for accessing the product and possibly fulfilling the need subject to access controls.
Travel
Hotel Booking Assistance
Generally, this is a manual search, comparison and booking task today. Sometimes, it is even limited to associated hotel partners and, in some cases, unable to support straight-through processing (STP).
Narrow AI - AI-based platforms could suggest hotel options based on customer preferences, generate automated recommendations and comparisons and transition to enhanced personalisation and deeper insights to guide the selection of accommodation and augmented services or activities. Depending on the domain and its access rights, it could see STP, finalising the booking and even through to the check-in, payment and check-out phases of the experience.
General AI - AI-driven search and hotel booking, AGI could plan entire trips, suggest tailoring experiences across the whole lifecycle, and have open choices of providers and services and source data to validate experience based on rating data from open sources. Basically, it's a comparison site on steroids.
Language and Cultural Understanding
Today, when an agent is connecting with a foreign service provider, they either hope the provider speaks their native language or that they have the ability to operate at a business level of proficiency to communicate needs or rely on language apps for translation.
Narrow AI - AI-aided language translation in real-time, both vocally and in written form, as it matures, it will have more nuanced industry-specific language skills and cultural insights, elevating the experience for both parties but will be limited within the trusted network of the agency.
General AI - AI-powered language translation both vocally and in written form of communication adapting at pace. Similar to Narrow AI it will expand to offer improved language skills, adapt to detected dialects and provide cultural insights. Its evolution will happen at pace due to it being exposed to more instances of it being used.
Retail Supermarket
Inventory Management
Today, it may be manual tracking and restocking-based rule-based metrics, actionable reporting and possibly some level of automated workflows across aspects of the activities.
Narrow AI - could use historical trends to predict demand forecasting and optimise inventory in real-time based on transactions at the point of sale (POS). It could mature to advanced analytics for precise demand forecasting and optimised inventory levels at both an organisation and local level and manage the process from warehouse to shopfront. Improving stock level accuracy and availability, look at feedback on Apps that complain about stock not being available at the POS. Access to buying will be limited to its domain or access levels to stock partners.
General AI - data-driven inventory management systems that would operate across the entire ecosystem. It could assess local and global availability, quality and logistics. As it optimises, it could revolutionise and drive efficiencies across supply chains and inform patterns to optimise buying and forecasting in real time whilst optimising efficiencies across the entire ecosystem.
Customer Service Automation
Serviced by rule-based workflow process management or supported by Chatbots with limitations across the services it supports based on ease or prioritisation efforts of the organisation.
Narrow AI - AI-powered Chatbots handle customer enquiries based on scripted responses or actions, as it matures over time, it could provide improved understanding and personalisation recommendations to customers based on purchase history, for example, sending a customer nudges about products they buy regularly that will be on special, prompting return visits providing tailoring and personalisation to support access and procurement of products and services. It could also be a predictor for inventory management within the domain.
General AI - Provide options to access products and services based on preferences and availability sources locally or globally across all active providers. For example, you may need to buy a present or fulfilling your shopping list. Over time it will predict with an understanding of usage patterns prompting when items need to be replaced or considering future purchases. With the maturity of smart homes, it will have the capability to integrate further with things like a smart fridge that can detect usage patterns and provide human-like interactions that are highly personalised as it develops a deeper understanding of the preferences of the people living within the household. IE You start to investigate the Mediterranean Diet, it could assess what is in your fridge or pantry and let you know if you have what you need to get started or what you need to source. It may access your calendar and alert you to plan for a special occasion, i.e. make a birthday cake, and it will let you know what you need to order.
I am glad to be able to share my thoughts and hypotheses in this domain, and look forward to hearing yours. I think it’s an interesting space that has already started shifting how we think and approach solving customer needs.
In upcoming posts, I will go deeper into Natural Language Processing and what to consider when thinking about your AI strategy.
If you are interested in my advisory or strategic services or have any questions, please contact me.
