AI is rapidly transforming the way businesses operate, streamlining processes, and enabling more informed decision-making across a wide variety of industries.
From automating repetitive tasks, to offering predictive insights, AI is redefining the meaning of efficiency, productivity and accuracy within business.
At our recent Data and AI masterclass in partnership with the Royal Institution of Chartered Surveyors, Professor Manuel López-Ibáñez delivered a thought-provoking session exploring the ongoing use of AI across the business landscape, and the questions leaders should ask themselves before they truly take advantage of AI technology to enhance their business activities.
This explorative session ignited various discussions about the future of AI, and today we’ll break down the key discussion points from the event.
Real world applications of AI
Most of us can probably think of at least one aspect of our day-to-day roles that could be streamlined to make our processes more productive, and in a number of different industries, AI is already helping to do just that.
Mathematical optimisation is one of the most mature branches of AI, with numerous business applications in planning, scheduling, logistics and manufacturing, among others.
The goal of optimisation is to find solutions that maximise (or minimise) one or more objectives (reduce costs, increase client satisfaction, reduce waiting times, increase fairness, reduce wastage, etc.) while satisfying constraints such as budgets, available resources, legal requirements, etc.
One example is the distribution of tons of food donations to hundreds of charities, which is a task that a large food bank must accomplish every day. Optimisation algorithms assist volunteers in allocating food donations based on fairness and waste reduction, while simultaneously planning the routes to deliver the food with the limited fleet of vans available to the food bank on the day.
Another branch of AI that’s increasingly finding its way within business practices is Deep Learning (DL), which involves training models known as artificial neural networks to learn from vast amounts of data. Large DL models are the foundation of recent advances in AI, such as ChatGPT. However, nowadays smaller DL models can run on a conventional computer.
One recent application of this technology is in the HGV industry, where companies responsible for transporting vehicles have to decide whether a client’s request can be fulfilled in the next trip without having to unload and reload any other vehicle, which may damage the vehicles. Deep Learning can make this decision in milliseconds, even before actually planning the trip and vehicle positions in the truck.
Training AI models
Once you’ve identified specific tasks or processes where AI could add value, it’s important to think about the methods in which to gather and organise data effectively in order to train AI models to recognise patterns, predictions, or automate tasks.
Training AI technologies requires businesses to take a structured approach to data collection in order to effectively integrate AI into their operations.
During the training process, it’s common for companies to go through data labelling, which often requires human input, as well as data cleaning to ensure missing, incorrect or irrelevant entries are removed.
It’s crucial that AI models are continuously monitored to detect any changes in data distribution, as well as retraining them should any model updates become available, which is especially important where customer behaviour or market conditions are likely to change.
The complexities and challenges of AI integration
Although implementing AI into operational tasks has proved to be a huge driver in productivity and efficiency for many businesses, it also comes with its own unique challenges.
Training AI models can be resource-intensive and costly, and requires significant power, storage, and specialised expertise in many cases.
Bias within AI can also occur when models learn patterns that reflect existing prejudices or inequalities that are present in the training data. This can cause huge problems leading to unfair or discriminatory decisions being made.
Data privacy and security can also present itself as another challenge, especially to businesses that rely on vast datasets, sensitive and personal information.
When AI models are trained on such data, there is a risk that the model inadvertently exposes personal information, especially in scenarios where the model's predictions or outputs include insights based on sensitive data, which could lead to copyright related and regulatory penalties further down the line.
Will AI work for my business?
There’s no doubt that the use of AI within business is set to dramatically increase within the next decade, with more and more leaders seeing the huge positive effects it can have to productivity, accuracy, and efficiency.
AI is by no means going to replace the whole workforce and there will always be the need for human expertise, but companies that delay their exploration of the potential benefits of AI implementation risk falling behind their competition.
So, if you’re considering taking the plunge and utilising AI, here are the questions to ask yourself:
- Do I need to provide additional data to fine tune my AI model?
- What are the security risks?
- What are the legal risks?
- How much training will my staff need?
- How much will it cost to maintain the system in the future?
If you’re looking to improve your ability to interpret and critically analyse business data and AI models to meet the needs of your organisation, join our 4-day Data and AI for Leaders course led by our academic team of AI experts including Professor Manuel López-Ibáñez.
