The promise of AI and why we’re not there yet

You know the story already. Artificial intelligence (AI) and the machine learning (ML) engines that drive it are about to reinvent every aspect of business and propel us forward into a new era of intelligent business systems, ultra-connected networks and sentient machines that can help manage and direct our workloads for us.

Except it’s not quite like that… not yet at least.

AI is arguably already the over-talked technology subject of the decade in spite of its undeniable second era now showing us real applications for the real world — it’s first era being the fanciful stuff of Hollywood Sci-Fi movies throughout the 1970s and 1980s.

So why isn’t real business really running on real AI yet?

One of the main stumbling blocks is business tradition, that is – all too many enterprises will do ‘things’ the way they’ve always been done because those systems just kind of work. These same firms run with a ‘why fix it if it’s not broken’ mentality.

Thinking about things

But what things are those business ‘things’ these firm’s won’t change?

They’re things like manual labor-intensive processes such as standardized report writing, financial record matching and all manner of business decisions based upon those oh-so valuable commodities of managerial experience and human instinct.

This old way of doing things practice needs to be broken. Even if it’s doesn’t need fixing, it does need reinventing. We need to wake up the internal divisions of business to make them realize the value in AI-driven intelligence if applied correctly.

This is the opinion put forward by Pedro Arellano in his role as vice president of product strategy at Birst, a cloud-native Business Intelligence (BI) analytics and data visualization company that became an Infor company last year in April 2017.

Palpable new trajectory

Arellano has pointed to a ‘new wave of disruption’ in the Business Intelligence sector that he saw firsthand when attending Gartner Data & Analytics Summits held in Sydney, Dallas, and London this year.

The excitement over this trajectory was palpable, he asserts, because companies are now demonstrating their use of AI-influenced augmented analytics features such as Natural Language Processing (NLP).

“Today we see that enterprise users can start to use NLP with BI and analytics tools to build reports without a mouse or drag-and-drop. Instead, users can now type or say,  ‘revenue, by quarter, by product category’ for example.”

“Natural language recognition enables these instructions to be declarative or more interrogative, such as, ‘What were last year’s sales by quarter, by product category?’ This new way of work will change business forever,” said Arellano.

Without deliberately naysaying and dragging down all the positivity on show here, the problem is that these types of work functions are being showcased at highly staged and choreographed (some would say contrived) Gartner analyst summits.

Crossing the chasm and getting to a place where this kind of thing happens in real business is another leap… arguably quite a big one.

Businesspeople are lazy, kind of

“While the above capabilities make for impressive demonstrations, they do not advance BI far from its long-standing function within the enterprise — that of building reports. Therein lies the problem; data analysts have the patience to build reports. Business users and executives do not,” said Birst’s Arellano.

What these same businesspeople really need, just by speaking into their smartphones, is answers to questions such as, how many units should I order, how many nurses should I hire, or what can I expect my revenues to be this quarter, by product line?

Democratization of analytics

So how can we move forward to a point where we truly embrace the democratization of analytics and bring the real promise of AI forwards?

The answer appears to lie in being able to show business departments where they can leverage BI for data beyond the data warehouse, automate data preparation tasks and interact with computers in a more natural, cognitive way.

Looking for AI tangibility then, let’s consider Clearlink. The company builds custom AI chatbots that recommends what to say to customers in real-time — and this is the differentiator. The platform allows for ‘hybrid interactions’ between a human chat agent and our AI chatbot.

Perhaps this hybrid element is fundamental to helping us cross the chasm. It allows us to humanize the computer side of AI just that little bit more.

“Our AI chatbot consistently learns with each interaction our chat agents have with customers. For example, the higher success rate certain agent chat responses have with customers, the more they’re going to appear.”

“Further, our AI chatbot even recognizes pivotal points in the conversation and offers ‘rebuttal’ responses in red when it detects the agent needs to shift the conversation in order to save the potential sale,” said Sarah Pike, communications manager at Clearlink.

This thread goes further, look at what Google is doing. ZDNet’s Larry Dignan this month reports on how the search and cloud giant is developing pre-packaged AI services aimed at specific business functions.

Google explains that its AI solutions fall into two categories.

The first Google AI set is its pre-packaged AI solutions that can be easily integrated into existing workflows.

The second is comprised of reference architectures that developers can use to create highly-customized AI tools that require more development work, but are suited to highly business-specific, integrated AI deployments.

“Google’s approach is worth noting because the cloud giants (AWS and Microsoft Azure) are likely to do something similar. The reality around AI is that there are specific industry use cases (think oil and gas, retail and manufacturing) but a broad set of functions that every company will need to use,” wrote Dignan.

It’s all about making AI more applied, more real, more hybrid, more tactile, tangible and altogether more touchy-feely.

Real AI applications are more human, who knew eh?