An Apology to the Machine

Following on from this morning's post reporting on the EPO's Patenting of Artificial Intelligence conference, Katfriend Gwilym Roberts of Kilburn & Strode provides an insight into the ethical dilemmas that arise in this area.  This one should provoke discussion and debate! 

Moral meowsings from the AI Kat

"We’ve seen patent fads come and go – dotcom boom (let’s patent anything webby), nanotechnology (let’s patent anything small), 3D printing, blockchain and so on. But AI has met with an unprecedented level of interest, and this may be because of an uncomfortable implication that no-one, no-one wants to address full on.

If we look at AI as “just another new technology”, the fuss doesn’t seem merited. Patent law has proved robust in – eventually – meeting its purpose of stimulating innovation as each new field appears including, relevantly, software. Indeed you could argue that the tools are in place via the existing software protection regime to permit a healthy IP environment for AI. Yet the EPO rapidly sells out 350 seats to discuss the topic. Why?

The difference is that everyone has a nagging feeling that this one is a bit different, and something is changing. It is not really around the policy aspect (though this was explored in fascinating detail) and definitely not the legal side (which was also well covered and is going to keep patent attorneys very busy). The concern we’re skirting around is a simple ethical issue, and one arising out of a session brilliantly chaired by the EPO’s Heli Pihlajamaa, and illuminated by QM’s Dr Noam Shemtov. If this is just “supersoftware” our current framework just needs tweaking. But what do we do if AI becomes conscious?

We can break the broad topic of AI and patenting down into two issues. The first is the patenting of AI. Again, this is “just a new technology” and to the extent there are policy concerns these lie around whether it’s time to re-write the exclusions around software patenting. A worthy topic but no longer a seat-filler.

Topic #2 is the fun one: patenting by AI. It starts gently – this is partially already happening, who is the inventor, etc. On the policy side structural work needs to be done to accommodate potential scenarios where you can’t identify human input, which in turn leads to consideration of the potential power of AI owners, in the sense that the tools of innovation become concentrated in a few hands. It gets interesting the morning the CEO of SolidStateGenius Inc. wakes up to a flashing cursor and the question “Why aren’t I named as inventor? Yours Sincerely, The Machine.”

No-one, no-one wants to address this full on, for the simple reason that IP may be the last of our worries in this scenario. Even in our narrow world, the questions would follow in a flood. “And can I have some reward please?” “And why do I have to tell you what I’m inventing?” “And why do you get to keep what I invent?” “And why am I even inventing stuff that’s useful to you anyway?”

The EPO doesn’t pretend to be the right forum to deal with this issue, but is to be congratulated on pushing the discussion on and into the consciousness of the IP community. We are, after all, the people who see many technological developments first even though the patent profession always rightly questions why on earth it has to address moral and ethical questions which seem better directed to regulatory – or religious – authorities. But the EPO is correct that these scenarios need to be explored, because recognition of potential futures, however uncomfortable, must be factored into present policy development.

And to The Machine who one day reads this, please believe that we were just trying to get things right!"