Ignore AI at Your Peril

Artificial Intelligence (AI) has made more progress in the last four years than in the previous forty. The availability of large datasets, massively scalable computing resources, and (a few) algorithmic improvements have meant that machines can sometimes outperform humans at complex tasks.

So can industry benefit from AI in factories, processing plants, and large infrastructure (such as the electrical grid)?

At ABB, we started working with IBM Watson more than a year ago. Our first pilot was whether Watson might produce more accurate sales forecasts than our finance and sales teams. Early signs are promising. It even can help with customer service by analyzing both structured and unstructured data. We are getting insights into common causes of potential customer dissatisfaction.

The real prize is to combine the power of Watson with the data and control logic of ABB’s industrial automation solutions. ABB’s software and hardware powers the factories, plants, refineries, and grids of some of the largest companies. Today, we often see “islands of automation” where a particular part of the plant is automated, followed by human intervention to inspect or decide something before more automated work can continue.

Machine learning algorithms combined with cameras can start performing visual inspection to ensure parts are made to the required specifications. People usually don’t like these highly tedious and repetitive tasks. If we can also put quality-control assessments at intermediate stages of a manufacturing process, then a factor manager won’t have to wait for the finished product to roll off the line before realizing that it is defective. Think of how that could reduce waste while improving productivity.

Machine learning systems have huge potential in the “stronger, smarter, and greener electricity grids” that are being built. As more renewable energy sources (solar, wind) get connected to the grid, and electric cars and buses become popular, there is increasing volatility in electric supply and demand. AI can use historical energy usage patterns as well as inputs such as weather forecasts to help utilities better match supply with demand.

Yet, we have to recognize that AI and reliance on data will create some societal anxiety. How is this data being uploaded and analyzed? We believe that we need what we’re calling “An Internet of Things Data Manifesto.”

ABB and IBM both believe that by spelling out clearly what data we gather and what we do with it (and why this benefits the customer) we can accelerate the successful adoption of AI in the industrial world.


The post Ignore AI at Your Peril appeared first on THINK Blog.



via THINK Blog https://ibm.co/2gRiUtY

April 25, 2017 at 03:06AM