How Organisations Can Create Impactful Outcomes with AI
In what ways can AI have a profound impact on your organisation?
Background
I was recently part of an “AI Day” panel discussion with a local university. The audience consisted of the university’s leadership team and senior faculty members. I was asked to share my thoughts on where I saw the impact of AI across the industries.
I’ve seen many organisations frantically jump onto the AI bandwagon because of the buzz that Gen AI has been generating. Nothing has caught management attention quite like this new technical innovation. These leaders are setting unrealistic expectations on the economic returns from this new tool, largely because they don’t understand how to contextualise AI for their respective organisations.
Now, I have fairly strong opinions about AI, often contrarian (not surprising if you’ve been following my articles here!). For example, I don’t subscribe to the notion of “ethical AI” and “explainable AI”. These are just nonsensical red-herring topics meant to make data illiterate management leaders feel they are part of the conversation. So in a similar vein, I’m going to deconstruct the AI phenomenon in the context of business transformation, and point out what should be obvious and what are myths.
And so I dedicate my 23rd weekly article to this topic.
(I write a weekly article on bad thinking and bad practices in data analytics / data science which you can find here.)
Simple Way to Understand AI
From a layperson’s perspective, there are currently 2 types of AI — traditional AI and generative AI (Gen AI for short). Traditional AI has been around for a long time; the term was first coined in 1956. Traditional AI is all around us today, embedded in our many operating processes and lived experiences. Traditional AI is simply the hyper-upgrade of ‘old-fashion’ statistical modelling using machine learning techniques and less structured data. Your credit score (used to evaluate your worthiness for a loan) is created using traditional AI. Forecasting your stock inventory replenishment is achieved using traditional AI. The AI that manages the health and charge/discharge of your mobile phone battery is traditional AI. Traditional AI solves problems related to decision uncertainty, and improves both the quality and speed of decisioning.
Gen AI burst on the scene only very recently, although research on the underlying algorithms started in 2014. The most famous Gen AI applications at the moment is ChatGPT (for text) and Midjourney (for visual). AI art is created by Gen AI. Modern interactive chatbots are using Gen AI. Gen AI solves problems related to information friction such as gathering and summarising information, or creating look-alike information.
Another simplified way to think about the difference between traditional AI and Gen AI is that the latter supports personalised human decision-making while the former is about making reliable decisions at scale.
Impactful Revolution
Which paths and options should organisations pursue to maximise the economic impact of AI? Now, traditional AI will always trump Gen AI in terms of economic value creation and extraction; reducing uncertainty in decision-making will always be more important. But Gen AI has the ability to ‘democratise’ data/AI/digital literacy by giving employees and customers a more natural way to interact with the machine. In an increasingly digital and knowledge economy, reducing information friction can be a game changer to productivity and user experience. Gen AI has the real potential to excel as an intelligent assistant.
But ultimately, it is the integration of traditional and Gen AI that will yield the greatest economic value. Consider the following example. My car’s service and repair warranty recently expired. I had an option to extend, and the only information provided was a brochure showing the terms and conditions, such as coverage up to the next 100,000 km or X years whichever comes first. To make a truly informed decision, I would have to compile my car’s service and repair history and what I would have paid had I not been on warranty. I would have to pull out my driving history to estimate the likelihood of achieving 100,000 km within X years. But these are information that my car service provider has on record, and they could easily show it to me, including what-if scenarios, so that I can make the best informed choice. But obfuscation may be in the interest of the car service provider. In such an instance, I could leverage Gen AI to do that information summary for me by connecting to my car log via API.
Consider another example. The traditional AI in Apple Maps computes the fastest route to our intended destination based on traffic conditions. But it isn’t able to give us advice on on choices we could make go to improve the economics of our drive such as time when we should leave our houses, taking different routes, better off taking public transport in some cases, etc. Gen AI could assist in this regard. The integration of traditional AI and Gen AI could completely change the way drive and use our cars.
Conclusion
Organisation leadership needs to recognise the kinds of problems that each type of AI is designed to solve for, and thus, the business value that can be generated through its implementation. AI is intrinsically interwoven with business digitalisation, and its nature is iterative. You won’t get it right the first time, but that doesn’t mean that the opportunity isn’t there.
