Artificial Intelligence (AI) is without any doubt the new fuel for the modern economy carrying with it the potential of disrupting and transforming almost every industry and business sector, shaping the future of work. And as with any other fast-evolving technology, AI also carries some serious implications and challenges towards its adoption.
What is interesting to note is that aside from the obvious technical challenges, it is actually too often the many social, economic, and organisational factors that can impact the successful delivery of an AI solution but are frequently overlooked.
At Galvia we safeguard Enterprise projects through AI-powered conversations and insights. We are committed to the advancement of AI driven by the ethical principles of fairness, transparency and relatability, human-centeredness, and privacy and security.
Personally, I strongly believe in contributing to knowledge sharing and insights as we all commit to the responsible use of AI in our communities. I have gained enormously from the wealth of research published and made freely available on Machine Learning (ML) and AI so I feel a very personal responsibility to share my own learnings.
Over the past four years as a team, we have deployed AI in enterprise environments and a certain pattern of deployment challenges has emerged. I want to take this opportunity to share with our peers these challenges or obstacles and the approach we take to circumvent those.
Cutting through the AI hype
Understandably there has been a lot of excitement in the media and amongst large technology companies about the future potential of AI and in particular robots becoming part of our day-to-day living. It is great content for sure, but it can unrealistically build up client expectations for what is achievable with the current generation of AI, which can in turn lead to disappointment both for customers and end-users.
I advise opening the communication channels and setting expectations with the client from the onset of a project based on what is realistically achievable with given data and constraints.
It’s all about the data
Unlike traditional software, AI and more specifically Machine Learning (ML) rely heavily on the data and its quality. AI is only as powerful as the data you train them on. Lack of data, poor data quality, and access to required data is a common scenario in most AI deployments and can heed the model training process.
Do a data audit before any deployment and proactively raise any issues or concerns. Let the client know that the only way we are going to get value out of AI is by linking the clinical or business problem to the organisation’s overall strategy and making sure we have a rich enough dataset to train the machine learning models so it generates actionable insights.
The future of work
Tied to the hype of AI is the fear that robots are going to take over all our jobs. It would be naive and irresponsible to say AI won’t replace some jobs. However, resisting change rather than preparing for what is to come is potentially disastrous for business.
AI is a tool for humans to get better at their work, save time, be more productive, and focus on creative tasks rather than be timelessly preoccupied with boring mundane work. We must remember that our human capacity for compassion and empathy is going to be a valuable asset in the future workforce and there are certain jobs hinged on care, creativity, and education that computers just can’t replace.
The responsibility is on the Enterprise to evolve and create its own unique identity, culture, work style, and management/reporting structure in the future of work. Research is already showing that those that do gain a competitive edge.
In the spirit of transparency and community, I hope that by sharing our learnings other AI vendors can successfully complete their projects with Enterprise, and together we collaborate and create an AI community that is equitable, secure, and above all human-centered.