What I learnt after a year in an AI consultancy and how it can accelerate your understanding of AI.
Last week marks my one year in business development for a London-based AI consultancy. During that time, I have been challenged by the sheer scale of the learning curve presented by the artificial intelligence sector and the many concepts and domains such as big data, cloud, data infrastructure and engineering that form part of the ecosystem and journey towards AI. Needless to say, the experience has been exceptionally rewarding and a career highlight. However, there have been additional challenges presented by the joy that has been 2020, for after a few months into my tenure, whilst still finding my footing, COVID-19 hit and the economy and business landscape has been trying to find balance in the gale ever since.
Whilst targets are a core part of sales, I have always felt that a critical responsibility of business development falls outside of achieving revenue targets. That responsibility is to educate and empower customers to make informed decisions as to what services and solutions best address their needs and achieve their business objectives. Although my industry experience is surely not the most diverse and so I am limited in my ability to make comparisons, I struggle to imagine an industry for which this could be more important but also challenging than Artificial Intelligence. The field is vast, the subject matter a fusion of multiple disciplines – from philosophy and behavioural economics to neuroscience and computer science – and rapidly changing. What’s more, it comes with a reputation and promise that few technologies other than perhaps the internet and blockchain could lay claim to: the promise to completely revolutionise society, industry and our lives.
So how can it be that something so revolutionary and critical is so poorly understood en masse?
In his famous book and TedTalk Start with Why, guru Simon Sinek explains how great business leaders have inspired and built business empires. The secret? Start with your why – the articulation of your purpose or vision that is at the core of everything you do – and let the how and what emerge from this origin. The method is designed to create alignment between the brand and customer values, the result being a raison d’etre that when combined with a great quality product or service results in a brand capable of capturing hearts and minds around the globe. Regardless of what you think of them (and their why), it’s easy to identify brands such as Apple, Amazon, Google and Facebook who have found their why and combined it with a winning formula that they execute consistently and brilliantly (again, regardless of what you think about it). They are the empires of modern times, and it’s no coincidence these empires are also pioneering advancements in AI.
AI is a field that has captured the attention of millions with an endless stream of articles focusing on the ‘what’ and ‘how’ of the latest innovations and use-cases which have fuelled speculation on the potential impact on industry, economy and our livelihood. Still, many business leaders struggle to understand what AI means for their business and where the opportunities lie. Why?
Well, from my experience, exactly that. Whilst many business leaders are very aware of the advancement of AI and its growing importance in their future technological and commercial roadmap, many have yet to establish how AI relates to their organisations’ needs and objectives, problems and challenges. Our focus on technology and how it is being used has eclipsed questioning the potential relevance. Simon Sinek might say our obsession with the ‘what’ and ‘how’ has distracted us from the ‘why’.
After a year spent at an AI consultancy trying to wrap my non-technical head around the technology, I cannot say I am surprised. If anything, it makes sense given that a large proportion of the literature focuses on the somewhat sensational and waffly aspects of AI or specific use cases. There is minimal opportunity for those intelligent yet ignorant on the matter to build a solid foundation and framework upon which to develop their knowledge on such a complex and diverse subject. The effect is that whilst it’s both exciting and useful to understand new and novel applications, without a framework for arranging new knowledge it winds up feeling as though one is accruing a lot of potentially vital information without much context: rather like having a puzzle pieces without the picture on the box to use as a reference to understand how they are related or might connect.
My goal is to provide exactly that, a simple framework that establishes the foundational concepts necessary to build your AI knowledge. That way you can ensure any new puzzle pieces acquired while building your knowledge of AI can be readily assimilated and understood. To do so, I’ll be using the method so simply laid out by Simon Sinek to create my own what, how and why of AI.
Let’s Begin By Answering The Seemingly Simple But Evasive Answer To The Question: What Is Artificial Intelligence (AI)?
Consider this thought experiment: if you were to ask a handful of people ‘what is artificial intelligence’ or ‘try to explain AI”, what would you expect them to say? What might you respond if asked? In all likelihood you’d encounter a range of responses, each with unique merit. These might include a reference to areas of AI, such as Natural Language Programming and the unique capability they enable such as the ability of machines to understand human language, or references to the complex statistics and mathematics that allow AI to generate useful insights from data. Now whilst this isn’t incorrect and are all great points, it also doesn’t quite capture the true scope of artificial intelligence.
Which isn’t unsurprising.
In the public domain AI is commonly associated with technology and often thought of as a discipline of computer science or data science. In fact, there are multiple disciplines that have contributed to the development of AI including philosophy, mathematics, biology, neuroscience, behavioural economics and arguably – given that humans are the template we aim to not only replicate, enhance and integrate with our societies built and occupied by humans – anthropology. Each of these has provided the foundation upon which we understand, design and build AI as we know it today and each will be instrumental in shaping a future with AI as a force for good.
This helps put AI into context as a vast, complex and rapidly evolving field with several different sub-domains (such as machine and deep learning), each responsible for enabling unique capability. And whilst technology is the enabler of artificial intelligence with all its data, algorithms and models, there is far more packaged into it. AI is a multi-faceted and nebulous domain and to date, no universally accepted definition exists as the capability and potential are always changing. What’s more, your understanding of AI is likely to shift depending on your perspective; whether you’re building it, using it or benefiting from it. It’s no wonder many people struggle to put their finger exactly on what AI is or does.
Having said that, most definitions do tend to hinge on a central theme, that “AI is the ability of a machine to imitate human behaviour and with that ability, perform tasks that normally require human level intelligence”. At first glance, this definition may not appear to offer any real clarity. There’s no mention of any of the buzzwords we normally see associated with AI. Can this really be what all the fuss is about?
But pause a moment and think about the implication of that definition. What it means.
Artificial Intelligence is the pursuit of imbuing humans’ expansive intelligence, faculties and competencies within machines. Our ability to learn and apply knowledge into critical and creative thinking to solve problems. Our capacity to absorb information from our environment via our senses to understand the world and objects in it and assimilate that information into our existing schemas. Our peerless skill of conceiving a goal and applying our cognition to develop and implement a strategy to achieve it, in many cases combining our intellect to accomplish tasks that would be out of reach for any lone individual. This is to say nothing of our emotional intelligence or lack thereof. Our capacity for love, to appreciate and be inspired by beauty. And when both our emotional and rational intelligence fails us, our decisions can be undermined by our worst impulses.
If successful in this pursuit, AI would stand among humankind’s greatest triumphs in a long history of grand endeavours. This is why AI attracts the hype it does. This is why AI is almost synonymous with buzzwords such as revolutionary and Industry 4.0.
But the AI of today is still far off replicating the richness of human intelligence and many of the hopes and aspirations of evangelists. Although the raw processing power of machines has been effectively harnessed to perform a diverse range of repetitious and even highly complex tasks from automating back office processes to identifying cancer, artificial intelligence is still limited by its requirements for vast quantities of high quality, highly accurate data and its inability to readily generalise knowledge across concepts and tasks. We are currently in the age of what is known as ‘narrow AI’, where with considerable training machines can excel and outperform humans at specific and discrete tasks, but is hamstrung by a lack of flexibility, critical thinking and self-awareness. This is unlikely to remain the case. The tide of AI has risen quickly to subsume tasks and responsibilities that until recently were in the once thought the peerless reach of humans. Innovation is being pursued relentlessly by pioneers aiming to create a future where human and machine intelligence opens up doors to endless potential and possibilities where the implications of some are likely to have been thought through better than others.
Whilst that is a conversation for everyone as a citizen of a world being increasingly shaped by AI, it’s not the questions many of us are grappling with. Those are ‘what is AI’ and ‘what does it mean for me?’. Whilst I hope to have sufficiently answered the former, as for the latter, I think it’s too soon to say with any degree of certainty. What I will say is the impact on our personal and professional lives is likely to be profound but unique and hard to ascertain. Rather like the nature of intelligence.
For those who still find themselves grappling with what AI is and what it means to them, I offer a simple metaphor. AI is both the tools and the toolbox; an array of mathematical and statistical techniques applied to data in order to perform a specific task and accomplish a specific goal that results in a desirable outcome. In much the same way as a hammer drives or pulls nails, tools such as computer vision in the case of AI, are responsible for recognising objects and actions in their environment. What’s more, the advantages of a hammer rapidly drop off outside of its narrow range of tasks. You’d be hard pressed to use a hammer to screw in a screw, similar to how a computer vision tool would be woefully unequipped to recognise human speech.
So Now We’ve Answered The What Of Artificial Intelligence (AI), It’s Time We Had A Look At The How.
If you’ve found yourself perplexed by the question of what AI is, you are likely to be even more confounded by the question of how AI works. A very deep and complex topic, my aim is to provide the key principles for non-technical stakeholders.
From the previous article and a cursory knowledge of AI, we know that AI can be applied in a variety of different ways and used to perform different types of tasks. AI is both a set of tools and the toolbox and we must ensure we select the relevant tool for the relevant task. Whilst the context of its application may vary, as do the tools and data used to generate the desired outcome, there is a method that is relatively consistent across all applications.
But before I begin, I first need to establish top-level clarity and alignment on a few terms that whilst you may have heard of in relation to AI, perhaps have never made a great deal of sense. Those terms are algorithms and models.
An algorithm is a set of instructions designed to perform a task. In the context of AI, the algorithm applied over time is what enables a machine to learn how to perform a task on its own, without or with minimal human intervention. The set of instructions that is the algorithm dictate what functions are performed on the data supplied to the algorithm, which is also relevant to the task the algorithm is intended to perform.
A model is what is created by the algorithm as an outcome of analysing the supplied data; a representation of the patterns and relationships that exist within the data. Think of an algorithm as an intelligent entity and the data being the entirety of information and knowledge of its world. The model is its understanding of the world – interactions, correlations and causations – which can be generalised to interpret and understand new data to which it is exposed.
Now that we’re clear on those terms, we can understand how AI works. Which is in two different ways.
The first involved feeding data (the ‘input) that is relevant to a specific task or objective and where the relationship between variables contained within the data are known, into an algorithm which analyses the data in order to create a model. The model, which is a mathematical and statistical representation of the patterns and relationships within the data, can then be used to generalise the identified relationships to new data, where the relationship between the variables is unknown. In doing so, a prediction is generated as the outcome (often called an ‘output’).I appreciate the above may require multiple reads.
Put simply, AI involves taking data that you do have to generate data that you don’t have. The value being that the generated output is based on massive quantities of high-quality, clean and representative data that makes it valuable to inform decisions about some unknown or hypothetical circumstances or conditions.
The second approach is perhaps more intuitive as it’s very similar to how humans learn to perform a majority of their tasks. It’s called reinforcement learning. For tasks too complex to program the machine, the machine interacts with its environment and learns through trial and error. Those actions that progress the machine towards achieving its goal are rewarded and reinforced, while those actions that hinder or don’t advance progress are discouraged. Reinforcement learning is a popular technique in applications such as robotics, where the challenges of programming a robot to navigate through a potentially infinite number of scenarios is infeasible and therefore having the robot learn is far more efficient and effective.
This brings us to what I have come to think of as the ‘underlying principle of AI’. In order to solve a problem using AI, the task must be expressed in a form a machine can understand and the machine must be supplied with the necessary data (the input) to perform or otherwise learn to generate predictions (the output) that enable it to accomplish its objective. By reframing business challenges as prediction problems and focusing on the types of predictions that help make decisions that solve business problems, we remove much of the complexity and unknowns from determining what are key priorities to be pursued using AI.
If business leaders are successful in recognising this, AI has immense capability and can do much of the heavy lifting in helping humans achieve our goals. However, it cannot do everything. In order to leverage the power of AI, we must first define and align on the goal or task we desire AI to perform before it can be developed and implemented. AI is the tool and the toolbox; not the hand that wields it, the eye that guides it or the mind that conceives the idea and outcome. For now, this is still the responsibility and domain of humans.
This may sound intuitive and even common sense. However based on my experience it is not something many consider when discussing AI. There is something about artificial intelligence – be it the hype, technical complexity or vast and nebulous nature – which wrongfoots many people. Which is an immense shame because such a simple mis-step is hindering so many businesses from realising the potential benefits of AI. Maybe there is too much preoccupation with how others are using AI or how it might be used in the future that people don’t immediately think to consider what problems or opportunities AI may resolve that are most relevant to them. This is what and the how distracting from the importance of the why. Making business decisions based on those made by others is not a robust long-term strategy. Particularly when, much like human intelligence, the ability of organisations to take advantage of AI is not distributed equally. This will ultimately come down to the core ingredients required to derive success from AI – clear objectives, quality data and viable technology – , the synergies between them and the standard to which a strategy to exploit the opportunity is planned and executed.
This is the concept I wish everyone understood about AI, before what it is or how it works. Why? Because broad, vague and unqualified statements about wanting to use AI to ‘reduce costs’ or ‘drive growth’ are obscure to most humans and utterly incomprehensible to machines (and the engineers that must program them). Making these types of statements as the basis for pursuing AI is likely to result in not progressing far, underwhelming results, wasted investment and missed opportunities. Now, to be clear: AI can achieve all these things and more. However it does so not by setting out to accomplish those vague objectives specifically but rather as a byproduct of excelling at tasks which cause those benefits to occur as a result. Examples of these tasks vary according to industry and application, but the underlying theme is the automation and optimisation of business operations, processes and workflows and informing high-value decisions made by business leaders.
So how can you benefit from AI? Begin with what you know the most about – (hint: not AI) – your business. Understand your goals and objectives, the challenges and obstacles that are preventing you from reaching them and the opportunities that, if harnessed, can help you achieve them. Data, algorithms, models and predictions might be how AI works but that knowledge isn’t what is key in helping you extract value from it.
Now that we’ve addressed how AI works – and more importantly, how you can approach AI to benefit from it – we have arrived at the most important question.
Why Should Your Business Use Artificial Intelligence (AI)?
As interesting as understanding what AI is and how it works, the most relevant and compelling question for any business is answering the question if and why they should use it. Whilst the specifics will vary for each business according to industry, vision and goal, there are some key aspects that I believe form the basis for every answer.
For any organisation, the foundation of the business case for AI will be based on three elements: the business objectives and the decisions made to achieve them, the predictions that inform decision-making and the value realised by acting on those decisions faster, more accurately and consistently over time than existing methods or by unlocking some new capability. This foundation must then be built upon, addressing factors such as data, technology that presents a viable solution plus the human and financial resources to develop and implement. By following this approach, business leaders will ensure they establish a clear business imperative that is agnostic and will form the basis for each of the other domains, allowing needs to direct AI initiatives rather than preconceptions based on technology or data.
The simplest themes for most organisations will involve cost reduction or revenue growth and applying AI to support this by optimising and automating a key business task, such as automating back office tasks, demand forecasting and inventory management or pricing strategy. Some businesses may investigate the new and novel applications of AI to enable new capabilities, such as personalised healthcare or to create music, art and film. Each of these cases are distinct and unique but what each has in common is that to someone they are high-value decisions that can be made more effectively with high-quality predictions. Part of the challenge organisations face in getting started with AI is identifying the types of high-value decisions they must make, how they should be prioritised and which are feasible to augment by introducing new technology.
To be clear, decisions do not refer solely to instances in which a human is actively required to make conscious, creative or critical decisions. Whether we choose to think of them in this way or not, organisations are made of decisions. Processes, operations and workflows are all comprised of decisions; be it hundreds or hundreds of millions, be they stand-alone or interconnected. For a given business, decisions may take place every microsecond across different platforms, applications and solutions all integrated together to form the complex technology lattice that enables the organisation to perform its services and create benefits for customers, whether an online transaction, money transfer or serving a digital ad on a website. These are all different types of decisions and as such, each requires its own unique logic to inform what decision should be made under different conditions and circumstances.
Currently, the ‘logic’ used by traditional systems and processes are explicitly programmed by humans into what is often an expansive and complex series of programmed instructions that essentially can be distilled into “if this, do that”. Humans program the rules of a task and the conditions under which certain rules should be executed and the machine performs them diligently. However, this approach has well established limits. There are many tasks which are too complex to program all the conditions and outcomes as well as tasks that we simply do not know how to program the machine to perform. These limitations have led to the rise of artificial intelligence to solve complex technical problems and the possibility presented by these technologies has captured commercial imagination and interest across industry from healthcare to marketing to automotive; with dreams of a future where AI enables early detection of disease and personalised treatment, where all marketing communications and offers are truly unique one-to-one and a world where self-driving cars become smarter the more they drive and are interconnected with other cars on the road resulting in fewer accidents and fatalities.The opportunity represented by AI is to introduce human intelligence into situations in which human logic and judgement is desired but human involvement is not feasible, scalable or otherwise limited due to some constraints on humans such as analytical skill, accuracy or speed. By embedding this capability in machines we slip the constraints of traditional human and machine limitations and are able to align human goals with the sheer analytical power of machines to overcome challenges and barriers that were potentially insurmountable unless the strengths of both were combined.
The power of AI is the ability of machines to make accurate predictions regarding outcomes that enable humans to better act in the future. Machines analyse what is known to ascertain relationships and predict outcomes that help humans anticipate the unknown. In doing so, machines help humans understand our world and vice versa. Although of high interest and value, humans lack the raw processing power to divine the depth of insights from data that machines, equipped with the right tools, can deduct far more readily (albeit they are far less adept at generalising learnings across tasks or context). Paradoxically, despite the efficiency and effectiveness with which machines may perform their task, fundamentally any and all information is arbitrary and presents no intrinsic value to machines as they lack purpose; they are devoid of the why. Well, outside of which is prescribed by humans.
Viewed from this perspective, AI is a partnership. Humans chart the course into tomorrow and machines help us navigate through the imperceptible fog that is the future. For now, the capability of AI is narrow; a specific tool for a specific task and humans must take the lead in deciding which is appropriate, based on our goals. One day this partnership could evolve into a form of symbiosis, with AI sharing our goals and becoming so intrinsic to our way of life that we can’t truly function or achieve our full potential without it, in much the same way as most people are likely to feel about modern technology. Perhaps even more compelling is the potential for AI to improve our understanding of how human intelligence works and what drives our behaviour. Replicating human intelligence in machines may end up being the greatest introspective experiment in human history that results in a new age of enlightenment and understanding.
But for now, this is postulation. The value AI presents to people and business right now is its ability to make better predictions that inform decisions to solve problems and achieve goals. To get started with AI, begin by setting aside the technology. Ask yourself what problems your business needs to solve, the challenges that need to be overcome and why this is important. Explore until you’ve hit the core of the issue, free from all the hype. That’s your why – and where your journey with AI begins.
originally posted on forbes.com by Kasia Borowska