This is a contributed piece by Ted Power, co-founder of Abacus
From Siri to self-driving cars, buzz has been building around artificial intelligence (AI). But Siri is far from able to answer every question or perform every task, and due to regulations it will still be years before most of us climb into a fully autonomous, self-driving car. In the meantime, there are many other products and services primed to be transformed by AI.
Machine learning is a tool, and like any tool, we need to learn how and when to use it. Machine learning requires large, structured datasets for training. And the types of problems that machine learning can help us solve need to be tightly scoped and clearly defined.
In the short term, our work lives are better suited to benefit from AI than our personal lives. This is because our work lives contain troves of structured data, and because our goals and intended outcomes at work are often more quantifiable. From scheduling meetings, to prospecting potential customers, to managing expenses, there are numerous applications in the enterprise where AI would be advantageous.
Where does AI fit in to business?
Imagine you run a business and you are trying to predict customer churn. A clearer understanding of which customers you’re at risk of losing will give you an opportunity to proactively message those customers and re-engage them.
You likely have thousands of rows of data for each customer (usage logs, support interactions, etc.). That’s too much data to read top-to-bottom, so today you are largely reliant on intuition and qualitative feedback.
Now imagine you could use your historical data to generate a customer churn model. The machine learning algorithm would identify patterns amongst customers that had churned in the past. Armed with this model, you could create a dashboard showing current at-risk customers.
This is of course just one of many possible AI applications. Think about how your sales team prospects potential customers. Or how much time you spend each week scheduling meetings. Or the manual effort required to manage cashflow and expenses. Or how the customer support team triages issues. In each of these instances, machine learning can help identify patterns and automate repetitive tasks.
Along with machine learning, natural language processing (a related branch of AI) will help us become more productive at work. Consider, for example, how organisations approach business intelligence. Typically information is pulled from an organisation’s database via structured query language (SQL) queries (or equivalent). But writing SQL queries is not trivial; there’s rigid syntax that requires deep technical knowledge. Imagine you could type in “how many iPhone vs. Android users were active last week?” as opposed to the corresponding complex SQL query. That would empower virtually everyone within your organisation to pull data and make more informed decisions.
What’s old is new
AI, broadly defined, is not a new concept. The underlying principles of machine learning have been understood for 50+ years. But over the past couple of years, AI has become far more effective at delivering real-world value.
Why has machine learning started working, seemingly all of a sudden? First of all, we now have better data: with cloud-based software, activity logs offer far more visibility into how people use our products. We also now have better hardware: computers are capable of the heavy lifting required to process all this new, multi-dimensional data. And finally, we now have better algorithms, as techniques such as deep learning are proving far more effective at extracting meaningful patterns.
Given these recent successes, machine learning is fast becoming mainstream. And people interested in harnessing AI should think about where there are opportunities within their organisations to leverage it.
Work work work work work
Effectively deploying AI requires a strategic shift in mindset. A lot of the more repetitive, rote aspects of our work will gradually be automated, freeing us up to focus on higher-value work. And as with any change, this will pose some challenges. Organisations will have to retool, and people will need to be trained to take advantage of these new capabilities.
But relative to the challenges we have faced adopting new technology in the past, getting comfortable with AI might not be so difficult. In many ways, AI enables a more intuitive relationship with technology. Instead of brittle interactions, interfaces will increasingly be more flexible and conversational. Instead of cursor and keyboard input, we will increasingly use voice and video as input (powered by speech recognition/NLP and computer vision).
From conversation to implementation
It’s one thing to talk about AI; actually implementing machine learning in the workplace is another matter entirely. Technology giants like Google, Microsoft, and Amazon are already battling over AI, and all other companies stand to benefit from that war, learning from and using what they create.
As the inventor Charles Kettering said, “a problem well-stated is half-solved.” This is particularly true of AI — if you can identify a tightly scoped domain, clearly articulate a desired outcome, and collect a training dataset, you’re halfway to a solution. Once an opportunity is identified, armed with the powerful machine learning frameworks from Google, Microsoft or Amazon, engineers will be able to deliver real value.
As the future of work evolves, AI will be a leading factor as it enables organisations to leapfrog legacy enterprise software, adopting a more modern tech stack. 2017 is going to be a big year for enterprise software + AI. Machine learning has officially left the research lab and is ready for widespread usage.
PREVIOUS ARTICLE«United We Stand – Nutanix CEO reacts to the Trump ban
Mark Chillingworth on IT leadership
Phil Muncaster reports on China and beyond
Kathryn Cave looks at the big trends in tech