Artefact UK’s VP of Data Ryo Katsuki sits down for a fireside discussion on the importance of incorporating AI into your business in 2020. He also advises on the most efficient way to do so.
Why now? Why is it so important to make AI a must for company strategy?
Thanks to technological advancements, high performance IT infrastructure has become so much more accessible to companies of all sizes in recent years. Companies now have fewer excuses for lacking facilities or methods to electronically record and use every bit of information.
The storage space is almost free, with Amazon’s AWS Glacier now offering data storage at #CONTENT#.000004 per MB, compared to 1967, when a 1MB harddrive, not even large enough to fit one of your iPhone photos, would have set you back million. In parallel, dollar per computing power has been decreasing by ten fold every four years.
Through cloud services, it has also become accessible to startups and SMEs who cannot cough up the large up-front cost of purchasing high spec machines or who lack the technical skills to maintain them. Companies can also further reduce costs. Most cloud companies are giving discounts to companies who estimate their future computing needs and are willing to reserve and pay for the cost up-front.
The opportunity here for businesses is obviously to capture all data, rely on computers to process it in machine-possible ways, and generate business impactful intelligence. To that end, cloud companies like Amazon, Google and Microsoft have been doing their part in developing turn-key machine learning platforms that are accessible and manageable for companies of all sizes. They now offer a whole library of machine learning algorithms to extract insights.
AI infrastructure, in terms of producing prescriptive narratives to help drive business decisions, is mature. It is now up to each company to ensure its business processes and talent pools are aligned to embrace the AI revolution. The companies that do not treat data as a core strategy and are not riding the current wave will not fare well against their competitors, especially those who manage their data well and propel business decisions to both generate and leverage artificial intelligence.
How easy is it to incorporate AI?
”It won’t just happen by hiring a few new graduate data scientists. It requires each company’s senior management team to come together and collaboratively relook at and redesign the way data flows within the organization.”
Good question. Incorporating AI is simple enough if you understand what it takes to make a change. That being said, incorporating AI into a business with real impact is a harder task than one might think.
I often ask business owners how extensively their company utilises AI on a scale of 1-10. They never say 10. It’s not just an IT department’s task. It won’t just happen by hiring a few new graduate data scientists. It requires each company’s senior management team to come together and collaboratively relook at and redesign the way data flows within the organization. It requires a data strategy, led with vision describing how the company would brand itself, market its products and services to its customers, produce and sell products, prepare and fulfil services, and operate internally to service internal and external clients.
All this is based on key default assumptions that all data will be collected and recorded electronically, that all business decisions will be made based on data or data-led predictions and recommendations. Business decisions that cannot be made based on data in reasonable time are funnelled into review for continuous improvement of business processes and data management. In the implementation, it is this continuous improvement program sponsored by all members of the senior management team that will play a key role in incorporating AI.
What would be the first steps in using AI in business?
Company executives could start by listing frequently asked business questions that continue to linger unanswered. These questions likely relate to every aspect of the business – the most common probably being ‘why is my revenue dropping?’ Positioning AI as a way to help answer these questions can form the starting point for an AI conversation.
Incorporating AI into a company is a cross-functional program, concerning all departments, requiring change management and system re-implementation. Ultimately, the senior management team has to come together and be ready to sponsor the AI revolution in the company.
If you are a data engineer or data scientist having a hard time pressing your AI agenda, find the business sponsor and decision-maker, and get his/her support first. Make sure you understand his/her business, present that you have plenty of imagination and be sure to articulate the benefits that outweigh the cost of change. At the end of the day, it’s a strategy and platform that you are presenting.
Don’t give up if you receive lukewarm support from one senior management team member. As you talk to other senior management team members, the cost, in theory, becomes shared, while the case for the benefits grows with support, making the cost per benefit for the company smaller.
What are the major opportunities with using AI?
”This illustrates the major opportunity of using AI in business terms: to reduce a colossal sized problem into a feasible, reasonable yet impactful problem.”
I’m most excited about using machine learning for early detection of problems that seem to come from nowhere or have specific causes that are hard to find, say, for example, paranoia. While there are lots of theories and different people will have different explanations for their own experiences, no one knows exactly what causes paranoia. Researchers may have identified some general risk factors but there are still so many that can be considered. Even in limiting the factors to physiological, sociological and mental, there are still more than 5 trillion to the power of 6 unique combinations of specific factors that can be considered. That’s astronomical.
This illustrates the major opportunity of using AI in business terms: to reduce a colossal sized problem into a feasible, reasonable yet impactful problem. In unsupervised exploratory studies, there is the chance to go wild with all the available raw data sources and identify factors that are previously unknown to be relevant.
For example, in the case of paranoia, to be able to say, using machine learning, that it is probable that factors such as sleeplessness have a stronger correlation to paranoia than other factors traditionally believed to have a stronger link such as childhood experiences, use of drugs, genetics, climate and etc. You could then prioritize the factors to monitor and identify early emerging signs to prevent some problems.
Today, AI and data leaders who run practical multivariate predictions and who use machine-learning techniques to resize impossible problems into practical ones are delivering real value. At the end of the day, we are still not ready to deal with the astronomically sized problems in the commonly accessible and manageable infrastructure.
Practically and technically, AI projects come down to business savvy engineers who understand business priorities, round up support, and design and implement data-driven heuristics and models that can deliver impactful intelligence. It’s the same set of business savvy engineers who can industrialize such intelligence, deliver such projects, and help implement AI-driven business setup.
If you’re looking to incorporate AI into your business strategy for 2020, get in touch with our team.