Overuse of ‘AI-Powered’ and What it Means for Organisations

Unfortunately, and often motivated by a perceived necessity to be part of the latest attention-grabbing trends, it’s not uncommon for organisations to jump the gun when it comes to joining the hype of a new product or technology – especially one that seems to sell well.

Now, I’m in no way suggesting that all companies selling products ‘powered by AI’ are making false or exaggerated claims, I’m simply asking customers to challenge these claims – particularly any that seem too good to be true. ‘AI-washing’ (as it’s sometimes dubbed) can lead to unintended consequences, not all of which have positive implications. For instance:

Hype and misleading claims 

The excitement generated around AI services (in books and films as well as everyday products like ChatGPT and Alexa) can lead to unrealistic expectations and some service providers overstating their AI capabilities or the benefits of their AI products. Ultimately, though, doing so only leads to dissatisfied customers and skepticism or disillusionment around the concept itself.

Misallocation of resources 

The truth is, not every technology problem requires an AI solution. Yes, AI and ML can be super useful tools. For example, when it comes to processing vast amounts of data, identifying patterns, making decisions with speed, and so on … However, thanks to the over-hyping of what AI can do and, consequently, the feeling that if we aren’t investing in AI, we are missing out, some organisations could waste resources on AI without having a clear understanding of whether it fits their needs or whether they will gain value from the product.

Security risks 

AI and ML systems, while offering significant benefits, unfortunately also introduce a variety of security risks that organisations must be prepared to address. These risks can stem from the inherent complexity of AI/ML technologies, the data they rely on, or the ways they interact with other systems and environments. Before implementing AI or ML services, then, it’s important to make sure that guardrails are in place so that the technology only interacts with data that is safe for users to access and that any systems and documents with privileged access are protected.

What’s the difference between artificial intelligence and machine learning? 

As mentioned above, it’s often the case that AI and ML get confused. All too often, this is done knowingly, as a job of branding or marketing. That’s because artificial intelligence sounds so much more exciting and sexier than machine learning. Indeed, ever since the likes of Siri, Alexa, ChatGPT, and Copilot came onto the market, the public’s interest in buying products labelled as AI has skyrocketed (and so have the profits of most companies selling ‘AI-powered’ products).

AI – referring to a broad field of computer science focused on creating systems capable of performing tasks that typically require human intelligence – has been around for a while, however. Its origins date back to the mid-20th century, in fact. AI’s rising popularity, driven by advancements in computing power, data availability, and algorithmic innovation, means it plays an increasingly significant role in our lives these days.

In a nutshell, there are three categories of artificial intelligence:

Narrow AI
Also known as ‘weak’ AI this is AI built to perform one specific task or a narrow range of tasks.  These systems operate under a limited set of constraints and contexts, focusing on a particular function rather than possessing general intelligence or understanding. Examples include voice assistants, curated recommendations, self-driving cars, automated chatbots, and image recognition.

General AI
Also known as Artificial General Intelligence (AGI) or ‘strong’ AI refers to a type of artificial intelligence that can understand, learn, and apply knowledge across a wide range of tasks, exhibiting cognitive capabilities similar to those of a human being. AGI aims to perform any intellectual task that a human can do, adapting to new situations and solving problems in a general way. Examples of achieving general AI include diagnosing and treating diseases with a deep understanding of medical knowledge or providing personalised tutoring/education tailored to an individual’s learning style and needs.

Superintelligence AI
This refers to a form of artificial intelligence and intelligence processing technology that surpasses human intelligence across virtually all domains, including creativity, problem-solving, and social interactions. It represents a hypothetical scenario in which AI systems are not only able to perform any intellectual task that a human can do but do so with far greater efficiency and at a much higher level of capability. Potential applications for this yet-to-be achieved AI might be accelerating discoveries in scientific research, optimising global economic systems, addressing climate change, and enhancing government decisions/ resolving conflicts.

Despite what seems to be the case when dramatised by headlines, levels of AI research today  sit somewhere between narrow AI and general AI. We are yet to fully achieve general AI and are a million miles away from sci-fi level superintelligence (sorry!).

Back in the realms of our current reality, and much more likely to be used by organisations and tech-companies today is, in fact, machine learning. ML is a subset of AI, yes, but it’s not fair to advertise machine learning as simply AI without explaining the unique differences between the two.

For instance, ML is focused on developing algorithms and statistical models that enable computers to learn from and make predictions or decisions based on data. We use it for spam filtering, for example, to segment customers for marketing purposes, and to identify anomalous behaviours for reasons of cyber security, etc. ML learning helps us understand or capture human behaviour and learn from it; it doesn’t think, learn, reason, and problem-solve like humans, which is the goal of AI.

In short, AI aims to mimic cognitive functions while ML creates systems that can improve over time.

Some key differences between AI services and ML services

Scope
AI includes the study of intelligent systems including ML. ML is a specific subset of AI focused only on algorithms and statistical models that enable machines to learn from data.

Objective
AI’s objective is to create systems that perform tasks that usually require human intelligence. ML’s objective is to develop algorithms that allow systems to learn from data and improve their performance over time without being explicitly programmed for specific tasks.

Technique
AI uses a wide range of techniques, including rule-based systems, search algorithms, optimisation, and more. ML primarily relies on data-driven approaches, including various types of learning algorithms.

Final word

To reiterate, this article is not a rejection of or critique of artificial intelligence. Indeed, there are many important use cases for AI and the technology promises continued advancements and widespread integration across various industries and aspects of daily life – machine learning being just one.

Rather, the challenge I’d like to leave readers with is always to question any AI tooling they’re considering purchasing, whether they have a genuine need for AI/ML, or whether this is the work of marketing making them feel like they do? 

I’d also ascertain whether the company you’re buying the tooling from is genuinely offering an AI product? And whether they’re being upfront about the difference between AI and ML and therefore the scope/capabilities of the product they’re selling?

It’s my belief that any organisation that pushes back on this sort of questioning is likely to be disingenuous and lack integrity – and that tells you everything you need to know.