Artificial intelligence (AI) may conjure up far-fetched ideas of robot assistants, or perhaps an all-seeing presence like HAL 9000, the sentient machine in the movie 2001. But the likelier truth is that AI will come in the form of software running in your data center.
And it will be coming very soon: Research firm Gartner predicts that “smart machines” will have a widespread impact on business within the next four years.
In general terms it’s likely that AI will be able to help IT departments do their job – and help businesses be more productive – by ensuring that “processes get applied, stuff is accurate, errors are eliminated, and compliance is met,” according to Dr Stuart Anderson, a research fellow at the Future of Humanity Institute at the University of Oxford.
It will also be quite unremarkable, according to some. “We see AI as another piece of software, like an ERP system,” says Marc Carrel-Billiard, managing director of Global Technology Research & Development at Accenture Technology. “It will be another tool in the CIO’s toolbox, and it will need to be integrated in to the IT landscape and connected to legacy environments,” he says.
But increasingly AI will be bundled as a feature into existing systems and products, he adds. “ERP vendors are already starting to put AI into what they do, and we will see more solutions with AI baked in,” Carrel-Billiard says.
In fact, professional services firm Deloitte Global predicts that by the end of this year, more than 80 percent of the largest enterprise software companies will have integrated AI functionality into their products, and by 2020 it expects 95 percent of the top 100 enterprise software companies to have done so.
Demystifying the AI hype
To understand why, it’s important to be clear about what AI can and cannot do. Scientists have so far made little progress in developing machines that can reason, but over the last 20 years huge advances have been made in the AI fields of pattern recognition and machine learning.
These allow computers to spot patterns in data, as well as images, text and spoken words, says Nova Spivack, a venture capitalist and CEO of big data company Bottlenose. “That means that systems will be able to sense and interpret the world – detecting weak signals early in large amounts of data,” he says.
For example, Deloitte cites a company that has enhanced it systems logging tool with machine learning capabilities which groups related server events together to make it easier for an IT manager to identify developing problems or unusual computing trends that should be addressed on a real-time basis.
But the use of AI will go well beyond IT departments to others such as HR that support the main business, Deloitte says. For example, AI will help HR departments forecast which applicants for call center jobs are likely to stay in the post the longest, or which applicants to other posts may have the best cultural fit and perform the best.
The same technology could also predict when a target candidate currently working at another organization might start seeking a new job – and make recruiters within the IT department aware of this, Deloitte suggests.
In a business context, AIs could also be used to help formulate tactics – for example working out the optimum amount to bid to acquire another company.
In order to get to work, artificial intelligence systems need data to learn from, and in the past it has been hard to provide them – and data analytics software more generally – with access to the many data silos in a typical organization.
But Mike Gualtieri, a principal analyst at Forrester Research, believes that the trend towards big data applications has largely solved this problem. That’s because companies are increasingly creating data lakes in Hadoop – making it for AI systems to access vast amounts of corporate data in one place.
Listening … and learning
AI may also be used to collect data within an organization, using a technique called “machine listening,” says Ruben Ortega, CTO at the Seattle-based Allen Institute for Artificial Intelligence. This type of AI technology has already been demonstrated in the consumer market by Amazon’s Echo device.
An enterprise machine listening device would record, process and “understand” words spoken during internal meetings and telephone calls. It could subsequently provide answers to factual questions such as “what budget was allocated to the project discussed in the meeting yesterday?” and perhaps even more predictive queries, such as “how much are our sales likely to increase if we increase our advertising spend as we discussed in the meeting yesterday?”
A particular area where the natural language capabilities of AI is already being used in an enterprise setting is in the form of “virtual assistants” which users can talk to (or interact with using a keyboard). For example, Carrel-Billiard says that a large oil and gas company already uses one to match employees with specific training courses that would suit them from the many that were on offer.
“Sometimes it could be a bit overwhelming for employees to find the right training, so we have replaced training managers with a virtual agent,” he says.
Fred Brown, CEO of Next IT, a virtual assistant software vendor, believes that this application of artificial intelligence within an organization is likely to be successful for a number of reasons. Firstly it provides accurate information and doesn’t overlook or forget things. Secondly, it can be cheaper than employing an expert to help people get information they can’t find. “Lower costs is what everyone is looking for,” he points out.
And increasingly, he believes, people expect and are happy to interact with a machine rather than another human being. In fact they may prefer that, he says. “Sometimes interacting with another person doesn’t feel right – there is a growing preference in society for people to go to a self-service channel.”
If the use of AI in the IT department sounds worrying, there’s no need to be too concerned about them running out of control just yet. That’s because of the difficulty that has been encountered in developing software that can reason.
“Very little progress has been made on that front,” says Nova Spivack. “That’s the dirty little secret of AI.”
And that means a sentient computer system like HAL 9000 is not even close to being on the horizon.
This article was written by Paul Rubens from CIO and was legally licensed through the NewsCred publisher network.