According to Declan Fay, VP of Strategic Business Development at Mobica, artificial intelligence (AI) implementations are prone to error and are vanity projects, expensive and do not deliver value to the bottom line.
These are all criticisms that have been levelled at AI since its early development and adoption. But, essentially, AI is just a natural extension of automation that has improved lives and businesses since the birth of the Industrial Revolution. The truth is there are huge benefits to be gained for companies that understand how to realise the value of AI. They key is understanding where and how it fits within your own organisation.
Start with ‘why’
One of the common mistakes made in AI implementations is a failure to define a project around its desired outcomes – what is the issue that my business is aiming to address?
Once you are clear on this, putting metrics in place that measure success becomes far easier. This enables you to demonstrate to all stakeholders, internal and external, the benefits that AI can bring to the business. Value should be seen on the bottom line, certainly, but organisations who view the opportunities purely through the lens of cost savings through human labour displacement are typically not the most successful.
Look at where you can measure value across the whole business; from efficiency or productivity gains and capacity utilisation, to enhancing the user or customer experience, ensuring quality and reliability to prevent unplanned downtime, improving worker safety and retention, or driving carbon neutral operations.
Identify where AI will work, and where it will not
Historically, automation technology was used for highly repetitive tasks that could be performed efficiently by large industrial robots, while tasks which were more dextrous or knowledge-based were reserved for human workers. However, AI-based automation means that many traditionally ‘human’ tasks can be performed just as effectively by a machine.
As a result of these advances, AI can contribute to so many aspects of a business that the question becomes more one of identifying tasks that AI should not do. Generally, AI is best suited to narrowly defined tasks, from which machine learning algorithms can be trained on a reasonably sized data set. Unsurprisingly, implementing data science and machine learning algorithms on the wrong type of data will rarely produce the results you are hoping for.
General purpose AI is a long way from reality. An AI algorithm may be trained to recognise faces and perform human emotion analysis, for example, but the same algorithm may not be able to recognise a cat or a bird. That’s why, essentially, the more complex a problem, the more complex the algorithm needed to solve it, and consequently the larger the data set you will need in order to produce accurate results. You need to be realistic about what you can achieve, and the effort required to get there.
Do not let AI become siloed
As well as the lack of a clear vision and unrealistic expectations, another pitfall when implementing AI projects, especially larger ones, is that automation and AI can become highly fragmented across the organisation.
Sometimes, instead of being a standalone initiative, AI deployment can be part of a broader digital transformation initiative, combining other enabling technologies such as Blockchain or IoT. For example, business process automation around tools like RPA (Robotic Process Automation), or customer interaction with NLP (Natural Language Processing) technologies will typically be driven out of a different parts of the organisation than automation for flexible manufacturing, or co-biotics in warehousing.
This lack of a holistic vision or orchestration creates a disparity between aspiration and reality and makes success much harder to measure. AI requires a more strategic project management approach to ensure that all initiatives deliver on the objectives set out and the wider business goals.
So, is it worth it?
AI is already delivering huge benefits to businesses. AI powered chatbots, for example, have made huge improvements to customer service by decreasing the time spent waiting for a human representative, while allowing businesses to use less human labour. Similarly, AI in connected factories can identify when components may be under strain or when a product is not being produced to the right standard – reducing downtime and wastage. Implementing AI can also help future-proof against potential disruptions, providing certainty to investors.
It is an exciting time for AI, and even if you think it is not relevant for your business right now, it soon will be. Successful implementations will enable machine human collaboration and automation of knowledge-based tasks like never before. But, in order for your business to deploy it effectively, you must have a clear vision of where it sits, a realistic understanding of what’s achievable, and a process for measuring and proving success across your organisation.