Will the AmeriKat be replaced by AI? (you can already get your cat robot courtesy of Petoi) |
"On 14 March aka Pi Day, Google announced that one of its employees, Emma Haruka Iwao, broke the pi world record. Ms Iwao used 25 virtual machines and 170 TB of data to calculate pi to 31 trillion digits (which would apparently take a human 332,064 years to say). The previous pi record was a mere 22 trillion digits.The AmeriKat wonders what the robots would say about all of this? But then, they weren't invited to the seminar... The slides of the event are here.
In light of this, it seems appropriate that the focus of AIPPI UK’s event, which was held on Pi Day, was on the use of artificial intelligence (AI) to assist with innovation, and the related legal issues.
John Brunner (Carpmaels & Ransford) chaired the event and was joined by a panel of experts; Dr David Brown (of UK start-up Healx Ltd and one of the inventors of Viagra™), Julian Nolan (Iprova Sarl), Fiona Stevens (GJE) and Jonny Moss (Hogarth Chambers).
The overriding take home point of the night was that machines are not yet taking over the world, at least when it comes to inventions. Or to put it less dramatically, machines are not yet coming up with their own inventions. AI will, and already does, assist in the innovation process. However, there is always a human behind the design of the algorithms, selecting data and then analysing the results of the algorithms.
Dr Brown and Mr Nolan discussed their use of AI to generate innovations. It was very clear that, in both of their experience, AI is used to assist and facilitate the innovation process, but we are a long way off from there being absolutely no human intervention. As Mr Nolan stated, AI augments and enhances the capabilities of inventors, but it does not replace them. It does not make the invention process any easier, just faster. For example, AI in the drug discovery process already facilitates processes that were routinely manual, such as the aggregation and analysis of biomedical information and analysis of medical records to identify clinical trial candidates, as well as identifying new pathways and targets (i.e., generating and identifying novel biomarkers to focus therapies).
Ms Stevens discussed the patentability of inventions which are created through the use of AI, particularly focusing on sufficiency and obviousness. One point of particular interest was whether, in due course, we may reach the point where AI generated data (in silico data) will replace in vivo data for the purpose of establishing plausibility. Further, if AI data is enough for sufficiency, will this have an impact on obviousness? At this stage, these issues are hypothetical but they may need to be considered in due course as AI continues to take a more prominent role in the innovation process. Ms Stevens concluded by echoing Mr Brown and Mr Nolan that AI is a useful mathematical tool, but there is a necessary element of human intervention in training the AI and analysing the data output.
Finally, Mr Moss rounded up the evening on the topic of inventorship (whether the AI should be considered to be the inventor or at least a joint inventor) and also whether the person skilled in the art should be considered to include the AI. Again, there were no definite conclusions to these issues, other than these factors will need to be addressed in due course.
In conclusion, AI is an important tool which will play an ever increasing role in the innovation process and there are a number of legal issues which will need to be considered to take into account the role of AI. However, AI is merely a tool used by humans and humans will remain as the ultimate innovators (at least for the time being). This conclusion is illustrated by the fact that the pi world record is attributed to Ms Iwao, not the computers. The computers used by Ms Iwao simply enabled her to break the record.
We can therefore all breathe a sigh of relief as we are not going to be replaced by robots…yet."
Next week there are two AIPPI UK events On Tuesday night, there is a blockbuster cast discussing the latest SPC developments including Charlotte May QC and David Rosenberg (register here). On Thursday night, is the Rapid Response on the Supreme Court's decision in Actavis v ICOS with some of the barristers who acted against each other in that case, including Tom Mitcheson QC (register here). These are going to be exciting events, so make sure to sign up as tickets are going fast.
What is called AI has nothing to do with intelligence as we know it.
ReplyDeleteIf a system is intelligent it should be able to go over and above the set of data which is given it to crunch. That the increase in computational power allows us nowadays to deal with large amount of data by finding a correlation between them does not render machines intelligent.
Machines will always only do what they are told to do. If the set of input data is changed, or if the correlation between data is defined differently, then the result will be different.
There could be some inventiveness in defining the set of input data or the correlation rules, but the raw result of the data computation will never be “intelligent”.
If a self-driven car has to choose between killing a pedestrian and killing the passengers of the car, this will be defined by the algorithm helping the decision taking, and hence defined by the programmer of the system. I fail to call this intelligent.
Neural networks are nothing new, and all the alleged importance given to AI is nothing more than a big hype. One more in the long list of hypes humanity has known.
Stop taking us for simple minds and define something as being intelligent which actually only is something defined by a programmer.