Machine rights

When an employee leaves one organisation and moves to another, they are not allowed to take the property of their first employer with them. That includes lists of customers, algorithms or other intellectual property. It doesn’t, however, stop employees from taking what they’ve learnt and applying it in their new role. The rules around what is fair (and legal) have developed over many years. We are just starting to explore the same questions now with robots powered by machine learning.

It is worth a reminder on the two main types of robot. The first, and the origin of the term, are those that manipulate the world around them supporting tasks like manufacturing, cleaning and an increasingly wide range of other real world physical tasks. The second are virtual agents that mimic real world user activity online, such as filling in forms, responding to emails and conversing on chat tools. Although the conventions are still forming, online agents are generally referred to as “bots” (derived as a shortened form of robot).

A debate on the role of bots (and robots more generally) moving between organisations isn’t academic, as most robotic process automation (bots replacing people in routine, often “cut and paste”, processes) are provided by third parties through the cloud. When a bot finishes with one organisation, what does it need to leave behind and what can it take with it?

There is no doubt that the data a bot deals with belongs to the company that created it. However, bots use artificial intelligence (AI) to get constantly smarter. The question is whether this AI-powered machine learning is deemed to be a form of data that is derived from the data that supported its learning.

It would be very easy to descend into a legal debate. My intention is to focus on what the right answer is to these important questions. Lawyers, guided by business, can then direct the development of contracts that support these positions.

If a business wanted to play to its own maximum advantage, it could insist that any machine learning done on their data was only to be used for their advantage. Taken to its logical conclusion, the consequences of such an approach would extend beyond bots to learning algorithms such as search engines. Search providers would actively resist attempts to isolate the activities of individuals in particular organisations from the constantly improving results they deliver for all their users.

Even if this position were possible to enforce, it would not be in any organisation’s favour unless they were the only ones that were applying such a rule. Any economy that allows the free flow of capability is better and more productive as a result. We all benefit by sharing as the machines we deal with get smarter.

However, taken to the other extreme, a robot that learns the secret algorithm behind the pricing or apportionment of a business should not be taking that knowledge to another organisation.

The difference, of course, between machine and human learning is the recall of the former. When a machine encodes something, it has total recall. By comparison, if a human sees a list of customers and their phone numbers their accurate recall would be close to zero!

The argument in favour of limiting machine learning derived from an “employer” would be that learning is at best an analogy rather than an exact analogue. The argument against is that everyone benefits as the pool of machine “employees” improves, a little like competing employers actively working together to improve the quality of professional education.

In my view, organisations overestimate the exclusivity or differentiation of their intellectual property. I also believe that they underestimate the power of working as part of a community that grows the whole economy. The most successful organisations grow the size of their market rather than treat it as a zero sum game. That doesn’t mean that businesses don’t have secrets that provide them with unique advantages, but rather that there are few that are genuinely valuable, they expire quickly and they are generally less valuable than having access to more capable people and machines.

Bots that learn across a community of businesses can actually make the whole economy stronger, no-one needs to lose in that equation!

Advertisements

About infodrivenbusiness

Robert Hillard is the author of Information-Driven Business, available through John Wiley & Sons. Find out more at www.infodrivenbusiness.com. Robert was an original founder of MIKE2.0 which provides a standard approach for Information and Data Management projects. He has held international consulting leadership roles and provided advice to government and private sector clients around the world. He is a Partner with Deloitte with more than twenty years experience in the discipline, focusing on standardised approaches to Information Management including being one of the first to use XBRL in government regulation and the promotion of information as a business asset rather than a technology problem. Find out more at www.infodrivenbusiness.com. The opinions expressed in this blog are entirely his own.
This entry was posted in business transformation, Digital, digital disruption and tagged , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s