Three challenges you need to overcome to thrive in the AI age

  • Posted on July 26, 2019
  • Estimated reading time 4 minutes
AI challenges

This article was originally written by Avanade alumn Maria Bines.

Read almost anything in the tech press today and you’re likely to be told that AI is the answer. The trouble is, I’m not always sure we’re asking the right question.

To demonstrate this point, nine out of ten senior executives believe AI will have a profound impact on their business and wider society. Yet half of them are struggling to apply it. So, there’s still plenty of confusion around what AI is, and what it can (and can’t) do.

Seizing the opportunity
That’s despite the fact that we can already show how AI can improve engagement with customers and employees, automate processes and analyse data to gain valuable insight. Companies at the frontier of AI are already reaping the rewards, reporting improved operations, marketing performance, customer service effectiveness and profits. In fact, Accenture estimates that by 2035, AI technologies will enable 38 percent profit gains.

AI has the power to do so much, but harnessing its potential requires organisations to get the right foundations in place and address three key stumbling blocks first – data, platforms and skills.

Challenge #1 – the data dilemma
It’s difficult to overemphasise the importance of data quality when it comes to AI. Just like analytics, the output of an AI platform can only ever be as good as the data itself. That’s why sufficient attention and diligence needs to be directed towards getting the right type of data (in the required volume and format).

If you’re expecting AI to be able to select, sort and cleanse your data, you’ll be disappointed. Well, you might be disappointed right now – though I can foresee AI actually being able to sort and declutter our data soon, identifying relevance amid tremendous volumes. Something similar to “the Marie Kondo method” for data.

But as well as data quality and quantity, there’s an ethical dimension too. Machine learning algorithms are written by humans, so bias can be unintentionally (or, more concerningly, intentionally) embedded into decision-making systems. Much of this bias originates in the type of data we’re feeding into AI algorithms and systems, so there needs to be a conscious effort to ensure AI is impartial and non-discriminatory.

Challenge #2 – the platform puzzle
AI is still an emerging technology, so it’s difficult to know where to place your bets when it comes to platforms.

Do you look to embrace a single platform, which can offer appealing simplicity and seamless interoperation with your current IT setup? Or do you opt for a more agnostic option, which will enable you to pivot as new approaches come along? In an ideal world, you would also be able to run multiple algorithms side by side across different cloud platforms with the same datasets –evaluating which model (and platform) is most appropriate for a specific business case. But it’s difficult to manage the code to achieve that.

Ultimately, most organisations I work with want to be able to adopt AI platforms but do so in a cost-effective manner (and avoid the need to re-architect their entire IT ecosystem). So, the predictable and somewhat awkward answer is… they want the best of both.

Good news, then, that we’re getting closer to cross-platform AI deployment processes. This will allow you to embrace new tech more easily and even read data in the cloud and on other platforms.

But, back to the platform dilemma I posed above – when it comes to the best bet for AI tech, I recommend hedging your bets. By that I mean finding a platform (or range of platforms) that give you flexibility today and the potential to connect and interact with more clouds in the future.

Challenge #3 – the skills shortage
Finally, it’s important to remember that AI is still an innately human process. While machines do a lot of the heavy lifting, AI is a partnership between your human workforce and your technology platforms.

That’s why a range of human capabilities are important – and it’s not just technical skills that are required for it to thrive. If we’re to make AI a universal success, a wide range of talents and proficiencies are required – everything from experience design, behaviourist techniques and psychology through to data science and software development.

We also need to ensure there’s sufficient diversity throughout that human involvement too – not just when it comes to gender or ethnicity, but experience, education levels and industry knowledge. By ensuring your human team is representative and balanced, you can develop AI that supports both customer and employee experiences effectively.

What you can do today
If this all sounds exciting but still feels distant, then let’s make it real for you. Here’s what you can do today to ensure you’re ready to thrive in the AI age.

Begin by making AI innovation part of your business-as-usual routine. And do that by creating an innovation strategy. It’ll give you a framework to take to your business (and customers) to identify processes that could work better. Or highlight services that are ripe for reinvention.

You can also use it to develop a pipeline of experiments – small projects with tightly focused outcomes. Your innovation strategy can deliver value quickly if you closely manage the scope, budget and timelines of these experiments. When you hit upon an idea that delivers impact on your bottom line, you’ll then have a readymade business case to support further investment.

We can help you to develop your innovation strategy and get ready to embrace AI.

Avanade Insights Newsletter

Stay up to date with our latest news.

Contact Avanade

Next steps

Talk to us about how we can bring the power of digital innovation to your business.

Share this page