The Internet of Things (IoT) is a prime example that highlights why data governance with respect to the quality and quantity of data is critical for successful machine learning (ML) and artificial intelligence (AI) initiatives. In fact, AI and data governance go hand-in-hand.
The large network of physical devices which make up the IoT is rapidly increasing. Gartner predicts that by 2020 it will include 20 billion connected devices, while IDC forecasts that figure will be closer to 50 billion, and says the data being created and copied annually will reach 44 trillion gigabytes.
Outside of industrial devices, data sources ripe for collection include emails, images, and video – alongside more everyday consumer items like fitness trackers, toys, automobiles, household appliances, and even the collar on the family pet.
Due to the business efficiencies and gains that can be realized through the smart use of information, it’s ultimately the coupling of software, algorithms, and intelligence with this gigantic wave of data that’s underpinning the advance of AI and ML.
Data makes for greater “precision” in AI and ML
Data governance for machine learning is critical to organizations attempting to build an AI strategy to improve their own products or services. John Fruehe, senior industry analyst, makes the point writing for Forbes: “Building a strategy on less-than-accurate data yields questionable results. It is important not to focus strategy on the products, technology, or pieces (things). Instead of focusing on the how of IoT, customers need to be focused on the what of IoT—namely the data.”
In a recent podcast series from TOPBOTS executive education called ‘AI for Growth’, Kevin Scott, Chief Technology Officer at Microsoft, echoes this strategic approach. He makes the point that data governance for machine learning is critical to understanding exactly what pieces of data an organization has or does not have for informing the types of AI it would be able to build.
In the podcast, Scott highlights two of the most interesting AI advances he has seen over the past year – developments in precision medicine and precision agriculture:
“With precision agriculture, we are entering an era where this intelligent edge, having these AI-capable devices everywhere including being able to mount them in drones, is allowing you to gather more interesting data about agricultural operations. The same thing is happening with medicine, where you take this combination of increasingly ubiquitous data about the human body that’s coming from smartwatches or fitness bands, then coupling this data with contemporary AI, like deep neural networks. The things you’re going to be able to do are really incredible, like predicting serious health conditions for virtually free before a patient is symptomatic, when it’s relatively easier to fix the underlying health condition than it is after the patient is sick.”
Data couples with language and human skills for conversational AI
Rachael Rekart, director of Machine Assistance for software firm Autodesk, was also a guest on the AI for Growth podcast. She led the development and implementation of Autodesk’s first application of artificial intelligence for customer engagement. Their virtual assistant, Ava, has reduced resolution times by 99% and cut costs from $15-$200 per ticket to under $1.
Rekart’s insights on the process of developing a successful AI conversational agent for customer engagement highlights the need for building the right kind of bridge between technology and human talent.
She explains, “Mostly when people think of (AI and ML) solutions, they think they need a data scientist and they’re good to go, but they’re so far from the truth! I have data scientists, I have computational linguists focusing on dialogue design and how to create or how to elicit a response through the way you’re phrasing something. I have creative writers, I have UX researchers, I have business analysts, I have communications managers. It’s really a lot of people that understand the value of conversation and how to bridge technology and humanities, because it’s such a blend of the two.”
For organizations seeking to implement a similar type of conversational AI solution she lists some useful milestones, which we’ve paraphrased here:
- Launch before you are ready and iterate often. Do not worry about achieving perfection immediately; instead, get your solution out there, get it learning, get it capturing customer inquiries, and then make sure you have the staff to iterate after you launch.
- Invest in talent, not just technology.
- Persona matters; put thought into how your company is to be represented. If you do not, your customers will.
- Be prepared for trade-offs, because you will find customers interact with you in all different ways. The industry moves quickly, so you need to be able to adapt and add new functionality such as image recognition, sentiment analysis, or all the other bells and whistles that enhance the overall customer experience.
What is your data strategy?
Given the importance of data governance for machine learning programs, have you developed a solid data strategy? Our experts can help. We have worked with many leading global companies that are today implementing machine learning at scale. Contact us today to learn more.