Artificial Intelligence is just plain dumb

Published on the 20/05/2017 | Written by Donovan Jackson


Google time machine

Google’s time machine reveals AI in service management is far from new…

Before getting too carried away with the contemporary frisson of excitement around machine learning and artificial intelligence, let’s just bear one thing in mind. It isn’t intelligent at all.

That’s the contention of Dave Wright, chief strategy officer at ServiceNow. “The biggest misconception out there right now is that AI is intelligent. It isn’t, because intelligence lies in the subtleties and that’s what sets human comprehension apart from anything that machines can do. When we get to levels beyond human comprehension, well, that will be real AI,” he told iStart at the company’s recent Orlando conference.

Like many other companies – from Google to Apple to Amazon and more – ServiceNow is looking at whacking AI into anything and everything it can. And why not – if AI and machine learning (ML) can automate routine tasks and workflows, more power to the workers of the world.

Much of ServiceNow’s AI comes from its acquisition of an outfit called DxContinuum. Its technology has been integrated into ServiceNow’s service management platform and is now called the ‘Intelligent Automation Engine’. The technology will evidently do clever ML things to help automate tasks for customer service, security, human resources and IT management.

But it’s not going to run the shop. Far from it, as the forthright Wright agreed when iStart asked if expectations for AI are going to exceed the results (again – and we’ll see why in a moment) and Wright said ‘Yes, in all probability’. However, “Where the big advances have been made is in compute power and data availability,  so it is the right time to take the next steps in AI,” he added.

Wright noted the two distinct types of ML: supervised and unsupervised (something touched on a while ago by Microsoft’s Chris Auld). “Right now, the capability is such that you have to show the machine what it needs to learn. That’s supervised learning – but most people think of unsupervised ML when considering the concept. That would represent genuine intelligence, and we aren’t there yet.”

AI is one of those industry concepts which tends to go around in cycles. Even as ServiceNow was talking about the introduction of intelligence into its service management platform, it brought back memories of the then-Computer Associates and its CA World conference held in pre-Katrina New Orleans.

Back in the Year 2000, the company now known as CA announced that it was releasing ‘software that can search through data sources to find patterns of activity and predict outcomes in the future…Neugents ii…can be used for virtually anything – from sales and marketing to network management to crime solving’.

Sound familiar? We’ve no idea what became of CA’s Neugents ii, beyond quietly disappearing (all Google searches for the term bring up links which are 18 years old. If anyone has insights, please share with us).

In any event, there is ample evidence of AL and ML delivering useful services to businesses and individuals today; you only need use your smartphone for confirmation of that. This also supports the contention that AI today is best at solving simple tasks – telling you the weather or providing directions.

Complex tasks are another matter altogether – but, as most people will appreciate, the ‘little things’ can get in the way of doing the hard work. Thus, said Wright in business as in consumer, “Expect AI to help automate the repetitive stuff, then move towards augmenting what people do to make them better at it. But without question, the ability to execute is getting there.”

Post a comment or question...

Your email address will not be published.

Time limit is exhausted. Please reload CAPTCHA.


Follow iStart to keep up to date with the latest news and views...



MORE NEWS: