Getting to Grips with AI Myths
It feels like everywhere you look at the moment there’s another article or event on the theme of Artificial Intelligence (AI). To say it’s a hot topic would be a massive understatement. Along with the hype, there’s a real sense of discovery - each person and organisation on a voyage to discover what benefits they can reap, and which hazards to mitigate or prepare for.
Myth: AI is only available to people with technical skills (data science, coding etc.)
Reality: A grasp of technical skills, particularly data science, can be helpful but any technologically literate person can use pre-trained models, or even experiences powered by AI, without having an understanding of machine learning.
Myth: AI is expensive and difficult to implement
Historically, AI has been expensive and limited to large and/or well-funded research organisations. Largely this is due to the infrastructure and skills required to gather, prepare, and store the data that’s needed to power AI. However, if your plan is to use existing AI tools rather than create your own, your organisation can benefit from AI without training your own model. AI services like ChatGPT are now available online so that everyday people can access them easily and at low cost.
Myth: Speed of output is the main benefit of using AI
Reality: Whilst AI can deliver efficiency gains in terms of streamlined operations through automation and improved workflows, it can also deliver other significant benefits. Clients that have used AI have seen additional financial returns from cost savings or greater personalisation of user experiences which delivers increased sales or heightened engagement. There are also possible non-financial benefits to be reaped - improved experience for clients and your team, greater creativity and morale, and the brand reputation associated with AI (often companies are seen as more forward-thinking and customer-centric)
Myth: AI is biased and unreliable
Reality: AI models are as biased as the data that they are trained on. If the dataset that powers the AI is biased, the AI outcomes will be skewed, biased, or perhaps even unfair. The AI is only as representative as the data that drives it. To minimise this risk, we would recommend you take time to understand how the source, collection, and processing of your data might have introduced bias and then use tools, frameworks, and adjustments to correct the bias (or at least highlight it for users).
Myth: AI is simple
Reality: AI is a broad field that encompasses all the ‘machines’ that are designed to perform tasks that would typically require human intelligence. It’s easiest to think of AI as an umbrella term that covers basic automation, advanced problem-solving, machine learning, large language models, natural language processing, and more. In its application, AI can appear simple - type in a prompt and receive a response for example, and refine accordingly but the software behind the scenes is highly complex.
Myth: An AI Strategy is only required for a tech business
Reality: For some companies, AI will open up new revenue streams, create efficiencies, and identify growth opportunities, for others, it will be a tool used by employees (perhaps even without the awareness of line managers) to expedite work or to develop design assets for example. Whether making your own AI tools or utilising those made by others, there are intellectual property implications that are worth managing through the creation of a strategy or policy.