Thursday, September 21, 2006

The Problem of George : Technology in Marketing

This year’s British Association of Advancement of Science’s annual conference in Norwich is an unlikely event of interest for marketing man, but a significant new prototype was being presented. Known as George, this is a conversational Artificial Intelligence [AI] system, designed for a contact centre environment, which can handle customer calls and log responses.

What is new about this system, reported, is that this one can track emotions, slangs and jokes. It can monitor the caller’s mood from the tone of voice and the use of words, and effectively escalate the call to a more specialised human agent in case it is not able to answer the query or the caller is getting annoyed. The system is designed to be adaptive too – learning from every call it could not handle, which it monitors and prepares itself with a response for next occasion.

The benefits of such systems are rather obvious. These systems, if effective, can effectively handle the first level customer interaction and can handle simple queries and responses. This will free up the resources for the ‘human’ team of customer service representatives, allowing the organisations to improve upon their quality of response by superior training and more selective recruitment, as well as by allowing them more time to spend with customers with complicated problems.

The scientists are enthusiastic that they may be able to develop quite effective conversational AI systems like George to be adopted in contact centre environments by organisations. Perhaps they are right. But, from a marketer’s point of view, adoption of such technology may still have three significant obstacles.

First problem is related to infrastructure. Systems like George are resource-intensive, and needs high specification servers to run the system. The scientists point out that the any actual implementation of such systems will be simpler, and a typical contact centre implementation may need only a few hundred thousand words – not the millions that the system has currently. Systems need computing power in line with their database size, and a scaled down ‘specified’ version will require less computing power, and the infrastructure can grow with the system, if needed. However, it is unlikely to be a simple proposition. A commercial product will need to follow the commands of economics as much as it does of science, and the ‘scaling down’ for economic reasons may not match the ‘scaling down’ perceived by its makers, and render the system dumb and ineffective.

The second problem is economic. Conversational AI systems may become smarter, but they will have to perform at an economic level and compete with millions of English-speaking ‘human’ agents entering or realigning in the workforce. Not just India, Ireland and Philippines will give George a run for its money, even the workers from the decimated manufacturing industries in the west will compete with George on cost terms. Remember, such systems will only start at the lower level simple tasks [forget about the scaling down for a moment – but even without that, conversational AI systems are only meant for simple tasks to start with]. And, remember, yes, AI systems will get trained over time once they are put in the job, but so will the workers. And, yes, there is a challenge teaching an Indian worker right accent and etiquette, but surely it is not harder than teaching an AI system to read human emotions correctly.

And, finally, the third problem will be of integration. A truly beautiful scientific invention doesn’t always follow the demands of commerce, as we were trying to point out above. The other end of the same problem is that often, real life services are a combination of many elements, and even a beautiful system like George, handling one part of the whole chain of functions, can only be as effective as the weakest part of the chain, often the management aspect of things – human decisions under pressure from the realities of the stock market. Without going into what may go wrong, let’s make the point – George needs to be able to work with other applications, some of which may be ‘not so great’. In reality, such integrations are rarely a success.

I am not sceptical of George, or for that matter, the technological inventions affecting various aspects of marketing. I am in fact euphoric, and know for a fact that technology has changed our lives and trades. But, I must warn against the pure inventions, and have a feel how far they must travel before becoming a productive system. When I read the story of George, this is exactly what popped up in my mind.

No comments:

Popular Posts

How To Live

"Far better it is to dare mighty things, to win glorious triumphs even though checkered by failure, than to rank with those poor spirits who neither enjoy nor suffer much because they live in the grey twilight that knows neither victory nor defeat."

- Theodore Roosevelt

Last Words

We shall not cease from exploration
And the end of all our exploring
Will be to arrive where we started
And know the place for the first time.

- T S Eliot

Creative Commons License