Artificial intelligence (AI) is set to remain a red-hot growth area. Research from the International Data Corporation (IDC) estimates that by 2022, global spending on AI projects will reach $77.6 billion: nearly three times the forecast for 2018.
But as telco, media and entertainment companies press ahead with AI initiatives, significant challenges lie ahead. As you scale your AI programmes, ensuring consistent messaging and eliminating bias become two of the biggest hurdles that stand in the way of maximising return on your investment.
Nothing quite illustrates this like the growth of the virtual assistant. Countless companies have recognised the potential of AI-based assistants and introduced them to their online presence or to assist call centre operatives. But how can you coordinate multiple assistants to provide customers with consistent information that informs and entices?
Every extra step on your AI journey can increase complexity and make these targets harder to attain. And that means a top-down, overarching strategy will be critical to achieving your customer experience (CX) goals.
Virtual Assistants: At Your Service
Virtual assistants have become one of the most visible of AI tools, and the momentum shows no sign of slowing. By 2020, Gartner believes 25% of all businesses will use the tools—up from just 2% in 2017. When the IBM Institute for Business Research surveyed C-level executives and CX leaders, it found 44% of companies planned to embed virtual assistants in the next two years. That represents an increase of 132% from today.
The benefits are clear. Millennials, in particular, are more likely to engage with service providers through innovative technologies rather than traditional methods like post, phone or email. When consumers do make a telephone enquiry, even the best human operators can only handle three calls an hour on average. With a virtual assistant on hand, that rises to nine.
How Media and Entertainment Brands Lead the Way
Some of the biggest media and sports brands on the planet have spearheaded adoption. International news providers have developed AI-based assistants that enable users to enter preferences on the types of stories they want to read, and receive a personalised breakdown tailored to their interests.
Sports teams have developed similar assistants: fans can set their favourite team or players, and receive highlights packages and real-time updates on breaking stories. Other news outlets have placed virtual assistants within articles on complex topics to offer readers a guiding hand through the content.
Movie studios have been even more innovative. As well as using the solutions to promote new trailers and publicise screenings, many have launched virtual assistants in the guise of the lead characters of the movie, enabling fans to communicate with them by voice or text.
Meanwhile, telcos are using virtual assistants to help subscribers enter questions on their data usage, to look up fixes for common problems with their service, or to recommend new deals to prospective or existing customers.
Taking the Next Step – the Challenge of Scale
So far, so good: where’s the problem? Well, it comes when companies move from running a small number of virtual assistants to 15, 20 or more. At that stage, you need a shift in mindset. Rather than experimenting with individual assistants, companies need a mature platform to synchronize messaging across all assistants.
That way, you can ensure users enjoy a seamless experience when communicating with all the different assistants on your website. Implementing a platform-based approach will enable individual assistants to talk to each other and share information, and help you to run deeper analytics on your data. That way your assistants know more on customer preferences, usage and tastes and can offer more personalized engagements, increasing the chances of upselling.
Consumers want to feel their provider really knows them and can deliver tailored packages around their needs. The danger with running many disconnected virtual assistants is that your service lacks coordination: users can end up having to provide the same information to different assistants, or receive one possible upgrade deal for their mobile and internet from one virtual assistant, and a different price from another.
Bias – Make Sure AI Works for You, not Against You
One other major trap to guard against is bias. Virtual assistants—along with other AI-based models—will only be as good as the data used to train them. The most well-documented form of bias is racial, as machine learning algorithms are typically built by a homogenous group of white, male data scientists. But over- or under-sampling when creating models will equally yield skewed and counterproductive results.
Let’s say you train an assistant to recommend a new satellite TV upgrade plan or new TV series based on an engagement with a subscriber. If the data you use when building the model includes lots of cases of users wanting to upgrade but ignores those who were keen to stay on their existing plan, then you could end up believing your product is more attractive than it really is and annoying your subscriber with repeat communications.
That’s why we built IBM AI Fairness 360, a group of opensource software tools that enable you to assess datasets and machine learning models. The tools will not only detect but also explain the bias in your model, providing you with the insights needed to remedy the problem before it leads to incorrect or unfair outcomes.
Moving on from the next step
In conclusion having a top down strategy as well as execution for AI is critical. Virtual Assistants are where a lot of the focus is today, but what else is around the corner that could be critical to your customer engagement. There are many examples of new uses of AI such as improved digital asset management that now allows real time personalisation of your content. Finding these new areas, exploring and then building a strategic approach to their use will be as important as virtual assistants.
With consumers wanting to experience media and content in new ways, Telco and Media & Entertainment companies must find new means to engage with existing and potential customers with the goal of developing new markets, new revenue streams and new business efficiencies.
At this point, supply chain professionals might be getting a bit impatient with AI. You’ve likely heard how this technology is poised to revolutionise the supply chain. But as a supply chain professional, you’re not working in R&D; you’re responsible for P&L. The obvious question is: what’s in it for me and my bottom-line performance? […]
With consumers wanting to experience media and content in new ways, Telco and Media & Entertainment companies must find new means to engage with existing and potential customers with the goal of developing new markets, new revenue streams and new business efficiencies. Earlier this year, I joined George Bevir, Editor, IBC365, to discuss the ever-growing trade-off of […]
Notorious examples of bias in facial recognition algorithms have received a lot of adverse coverage this year. It highlights the explosion of bias in AI systems and algorithms, but according to IBM Research, only unbiased AI will survive. To counter such bias, companies like IBM have been making data sets and toolkits available. An example […]