Over a million developers have joined DZone.

Closing The Skills Gap That's Holding AI Back

DZone's Guide to

Closing The Skills Gap That's Holding AI Back

We're moving on from an age in which humans adapt to how computers work, to one in which computers adapt to us.

· AI Zone ·
Free Resource

Bias comes in a variety of forms, all of them potentially damaging to the efficacy of your ML algorithm. Read how Alegion's Chief Data Scientist discusses the source of most headlines about AI failures here.

Reading the news every day will subject the reader to a deluge of hype around artificial intelligence. Depending on the perspective taken by the writer, the technology will often either be about to revolutionize things for the better, or, more probably, for the worse. However, we should be under no illusion that it will revolutionize things.

With such a barrage of hype, it's easy to fall into the trap of thinking that our organizations are already being disrupted on an unprecedented scale by AI-based technologies. That the future so giddily predicted is already here.

Except that's not entirely true. A recent executive survey by EY revealed that the vast majority of organizations are doing little more than tinkering with AI. They're doing tests and pilots, but nothing more. What's more, the survey also painted a bleak picture in terms of the prospects for scaling these projects up.

Organizations not only lacked board level buy-in, even for AI deployments that improve what is rather than fundamentally reshape business-as-usual, but they're also struggling to find the skills to power their pilot projects, let alone anything more substantial.

The AI Skills Gap

Suffice to say, most of the time, when executives talk of a skills gap, they refer to the technical skills to actually develop AI solutions in the first place, but as Accenture's Paul Daugherty and James Wilson highlight, there will also be a need for new skills to capitalize on the opportunities presented by AI.

In their latest book Human + Machine, they outline three classes of jobs that are likely to emerge as humans begin to work more closely with AI.

1. Training

We're moving on from an age in which humans adapt to how computers work, to one in which computers adapt to us. In order for this transition to occur, however, we will need people to effectively train machines to work alongside us.

This could involve ensuring the data that machines are trained with is suitable, or correcting errors and reinforcing successes in machine behavior.

2. Explaining

AI systems are taking on ever more important roles, and as they do so, the ability to explain their working becomes ever more important. Therefore, being able to shed light on the black box workings of an algorithm will be key.

Humans may test, observe, and explain the algorithms or augment their interface to make them more explainable. It also seems likely that there will be a sizeable role for interpreting machine outputs, especially in areas such as healthcare.

3. Sustaining

There is also likely to be a significant role for people who will ensure that AI-systems operate as they should. It's crucial that AI serves us rather than the other way round, and people in this category will fulfill that role.

Tasks might include setting limits for AI-systems or flagging errors in machine judgment. It might even include a "machine resources" department in the same that we have human resources departments now, who will be tasked with assessing and evaluating performance levels.

The S Curve

It's clear, however, that these skills don't exist at the moment, and it's holding us back in terms of exploiting AI technology to the fullest. So, what can organizations do about the skills gap that exists, both in terms of explicit AI development roles and also the wider skills required to work with AI effectively? At the moment, there's something of a dichotomy in that whilst most organizations accept that a skills gap exists, there is a reluctance to invest in training staff to develop those skills.

We have a situation where the skills are largely not available in the labor market, yet employers aren't investing to develop those skills either. A recent study by Future Workplace and The Learning House highlighted the scale of the challenge, with huge numbers of roles unfilled due to shortages in the labor market, and with the vast majority of organizations investing paltry sums in training.

"With more than 6 million unfilled jobs in America, the skills gap continues to stall business growth and innovation. We believe that training the workforce for current and future skills will be the most effective and efficient way to fill the skills gap," says Future Workplace's Dan Schawbel.

To achieve that level of reskilling requires a fundamental shift in mindset. As Whitney Johnson says in her latest book Build An A-Team, the learning process follows an S curve of low productivity to begin with before rapid learning takes us to a sweet spot of maximum productivity.

Hiring for Potential

As the work of EY and Future Workplace suggests, most organizations today are hiring for the finished product rather than the potential that can be developed. They want new recruits to be at the end of the S curve, not at the beginning.

It's hard to imagine AI achieving its considerable potential until organizations get to grips with the need to invest considerably in the constant learning and development their employees need.

Whilst we are not at that stage yet, there are signs that there is change afoot in how we approach skills, both in potential recruits and existing employees.

Professional services firm Cognizant are seeing signs of this change through their work. Ben Pring, the founder of Cognizant's Future of Work Center, told me recently that there is a historical lag between the emergence of a new technology and the skills required to make it thrive, and we are in the midst of that at the moment.

He does believe however that both employers and governments are slowly making the changes required, and points to the emergence of MOOCs as a mechanism by which we can equip ourselves and our employees with the skills of the future.

We've seen an "X+AI" approach across industries, whereby AI is bolted onto existing products and services, and a similar approach is likely to be required in education too, with traditional courses requiring an AI element to them to equip people with the skills to apply their knowledge to the AI-driven world.

Cognizant recently launched a new $100 million foundation to help provide those kinds of skills, and they join the likes of Google and NVIDIA, who have both launched similar projects this year. In the grand scheme of things, these are relatively small steps, but they are at least a sign that society is moving in the right direction. There is a lot more that needs to be done, however, and we cannot wait much longer before stepping up to the plate.

Your machine learning project needs enormous amounts of training data to get to a production-ready confidence level. Get a checklist approach to assembling the combination of technology, workforce and project management skills you’ll need to prepare your own training data.

ai ,artificial intelligence

Published at DZone with permission of

Opinions expressed by DZone contributors are their own.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}