AI and the Year Ahead: What Now?

Covid-19 prompts new pragmatism in enterprise approaches to AI

As the world fights the Covid-19 pandemic, businesses are overhauling their IT strategies by continuing to prioritize remote working, business continuity and ensuring networks, infrastructure and security can cope in the new environment.

These efforts will remain central IT themes for the next 12 months at least. However, based on conversations I’ve had with customers over the past month, one technology is set to undergo perhaps the biggest shift of all: artificial intelligence (AI).

As a year of uncertainty looms, driven by lockdown periods and the high likelihood of a global economic recession, a new pragmatism is now characterizing approaches to AI. This will define customer AI strategies and market developments for the near future.

Let’s take a look at what’s changing in enterprise AI as result of the Covid-19 pandemic. I see several trends on the horizon carrying important implications for the market in a turbulent second half of 2020.

AI and the Enterprise Market: The Timeline

Firstly, let’s examine the timeline ahead of us. I don’t have a crystal ball, and many variables could change the timing of the crisis, especially in this early period. But based on my discussions with customers and colleagues at CCS Insight, I believe there will be three distinct phases. Each period will influence the level of technology and investment in AI that enterprises will make over the next 12 to 18 months.

Phase 1: Full Lockdowns

The current phase, for the most part, continues to see tactical responses by businesses as they scramble to get employees working from home. As the lockdown continues, people’s routines become established and processes start to standardize in the home-working scenario, and companies will work out which technologies work best in this environment. It’s why technologies such as remote collaboration, desktop virtualization, device management and security and VPN connectivity have been among the main focus areas so far.

When it comes to AI, many of the headline-grabbing announcements over the past few weeks have understandably centred on sectors that are vital to the fight against Covid-19, such as government and healthcare, and in helping them rapidly apply AI technologies like conversational agents to their response.

IBM, for example, launched Watson Assistant for Citizens earlier this month, a solution the company made free for 90 days to help governments, health companies and academic institutions handle enquiries. Microsoft, too, announced in March that the US Centers for Disease Control and Prevention is using its healthcare chat bot service — which became generally available on Azure in February — on its website to help assess symptoms of the disease and risk factors.

It looks as though this lockdown phase will continue for another month or two in the UK, possibly longer in the US and potentially shorter elsewhere in Europe and Asia, depending on the many variables associated with the pandemic, as well as responses by governments and companies.

Phase 2: Rolling Lockdowns

The next phase should kick in once the current full-scale lockdown starts to ease, schools start to reopen and pockets of people return to offices, most likely once there’s widespread antibody testing. Vulnerable groups will still continue to self-isolate and work remotely.

This is likely to result in lockdowns being reinstated for short periods, either regionally or more broadly owing to recontamination. This second stage will continue to be an unstable period, compounded by the added financial pressure that many businesses will be under as the impact of a recession becomes real.

The need to test, track and trace the virus on a vast scale as well as model outcomes during this phase will of course open up huge opportunities for AI, especially with governments. But when it comes to enterprise technology decisions, many companies will struggle to make large-scale or strategic investments during this period, triggering several important changes to the market that I’ll cover later on.

Phase 3: Post-Covid-19

Real tangible change to the current lockdown picture will only appear in this phase, which is based on the wide availability of a vaccine that allows protection of vulnerable people and free movement more broadly. This is widely expected to be at least 12 months away, and possibly longer considering the time it will take for a vaccine to reach everyone around the globe.

Additionally, in this phase we’re unlikely to see a return to the blanket office-based work environment that many organizations had before the pandemic, as remote working becomes embedded and many businesses will operate hybrid models. In my view, only at this point, roughly 12 to 18 months from now, will businesses return to taking a more long-term and strategic view on technology and AI investment decisions.

AI and the Enterprise Market: The Impact

So how will this timeline affect the fate of AI projects? Below are four major trends I expect to come into play. I believe these will be key aspects shaping customer strategies and market developments for the next 12 months.

AI Experiments (and Toothbrushes) Are Gone

First of all, it’s clear the halcyon days of AI experimentation and hype are over.

Late in 2019, a remarkable 75% of US and European organizations stated that they were using, testing or researching the deployment of machine learning in their businesses, according to a survey of senior IT leaders conducted by CCS Insight (see Security, Cloud and Ethics Dominate IT Priorities in 2019). But for the vast majority, AI has been an experimental, workbench technology. Most machine learning models built today fail to make it into production and remain in what some call “pilot purgatory.” Added to this is the hype for the technology, fuelled in part by suppliers “AI washing” their products over the past few years. At Mobile World Congress 2018, for example, we counted at least three companies claiming to provide the world’s first AI-equipped toothbrushes.

The change in climate means that companies can no longer afford to have significant investments tied up in longer-term projects and proofs of concept as many did in 2019. Explorative, innovation projects and hyped products will now be shelved in favour of initiatives (and suppliers) that can solve specific business problems in the wake of the pandemic. There will also be much more focus on solutions that deliver clear operational improvements and can get into production quickly.

Narrow AI and Automation Come into Focus

The need to quickly solve specific business problems as a result of the coronavirus outbreak will be a boon for more short-term, narrow AI projects that target automation, operational gains and enable process reengineering.

Narrow AI is a term used to describe AI systems that are designed to handle a singular or limited task or operation. Applied to operational areas under strain from the pandemic, technologies such as robotic process automation, conversational agents and process mining will ramp up more quickly than more complex, innovation-targeting projects based on deep learning, for example.

One area under enormous pressure during all this has been contact centres, as businesses shift employees to home working and customer enquiries skyrocket. More companies have been turning to AI, and conversational agents in particular, to help with their operational challenges in this area.

The market is also responding to this requirement. Google Cloud, for example, launched a Rapid Response Virtual Agent program on April 8 as part of its Contact Center AI solution. One of its customers, the Oklahoma Employment Security Commission, has been using the product to help handle the more than 60,000 unemployment claim calls it’s receiving every day.

Automation Anywhere, a developer of robotic process automation software, also stated this month that a large US hotel chain is using its software to build bots that can help its remote call centre employees reduce wait times and handle reservation cancellations, which have ballooned during this time.

Both are prime examples of the types of AI automation project we’ll see more of in the next 12 months.

Cost and Business Metrics Are Now King

Perhaps the most important change of all will be that enterprises will double down on projects that yield clear and immediate economic value, as organizations apply much more rigor to measuring business as opposed to technological metrics.

In the past, many AI projects were primarily measured on the performance and accuracy of the AI models themselves. This will now shift to business outcomes the AI is applied to such as revenue generation, output and improvement to customer satisfaction.

With less cash at their disposal, companies will also turn to AI to help them save money, especially in areas that are likely to go through rationalization, with examples being customer experience and marketing, supply chain and manufacturing.

Without the time nor headcount to wait six months for projects to deliver value, companies may also look to buy more complete solutions from providers rather than building them. I can also see companies focussing more on infusing business metrics into the MLOps process and development life cycle itself.

A Shift to Responsible Approaches

The fourth and final trend I expect to materialize over the next 12 months is the attention paid to responsible AI. This is a combination of principles, practices and tools that enable businesses to deploy AI technologies in their organizations in an ethical, transparent, secure and accountable way.

I’ve been asked many times during the past few weeks whether the heightened pressure enterprises are facing will cause them to short-cut aspects like governance, transparency and bias when building AI models in favour of getting them into production faster. This is certainly a possibility, but in my opinion, people’s memories of the actions that enterprises are taking now will run much deeper than many of the better-planned projects that have come before the pandemic or have yet to start. Customers will therefore aim to get AI right during the crisis as well.

A responsible AI strategy today will not only help to reduce potential costly problems later on, but above all, will engender confidence in the AI that gets built today. This will ultimately lead to faster deployments, wider adoption and most importantly, trust in the companies that adopt this approach, which will last well beyond the current crisis.

The Final Word

AI will remain among the tech industry’s most important trends over the next 12 months, as it had been prior to the pandemic as businesses require more from their data. The technology is finally being deployed and recognized as a force for good, which contrasts the fear and questions surrounding AI over the past few years.

But a new pragmatism is entering the market as businesses react, recover and advance beyond the current situation. Gone is the hype and widespread experimentation with AI. In their place is a more practical focus on projects that deliver immediate business results, improve operations and establish trust. Other opportunities for the technology will arise further down the line as a result of Covid-19, including in document understanding for changes in the compliance landscape, in robotics within manufacturing and in further scenario planning applications as part of expanding business continuity programmes.

C-suite executives are rightly focussed now on the health and safety of their employees, enabling customer engagement in this period and maintaining business agility, as events continue to unfold almost on a daily basis.

This is creating new opportunities and direction for the AI market. It will be fascinating to track how customers and suppliers respond to these changes over the coming months.