USD 7 billion patent lawsuit against Google - can Ambercite AI predict an outcome?
Ambercite has been used to predict and analyse a number of patent disputes in previous blogs, and results have matched the outcome.
With this in mind, I thought it would be interesting to analyse the $7b patent dispute between Singular Computing and Google, which is based on assertions that Google is infringing patents filed by Singular Computing in relation to AI.
Ambercite uses an AI algorithm to predict similarities between patent portfolios - so what does Ambercite say about this patent dispute?
Patents in question
The three patents in question are US8,407,273, US9,218,156 and US10,416,961. These are complex patents, but in short cover a fast processing device that is thought to help with AI data processing.
Long previous experience has suggested that if a company is interested in a technical area, they will file patents to protect this interest. So in this case, we could ask - according to Ambercite, are there any similar patents filed by Google to the three Singular Computing patents?
If so, this suggests that Google may be working on similar technology. If not, this would suggest that Google is not.
Now of course, by this time patent lawyers might be hot under the collar and say it is about claim construction, expert witnesses , behaviour by the alleged parties, legal rulings, jury decisions and so forth. And they would be 100% right , and we need to have this process play out accordingly. However, it still might be fun to run through our process, make a prediction, and then see what happens.
The basis of an Ambercite licensing analysis is the running of a patent search based on the above three patents. The patent input screen looks like this (Ambercite recognised that the three patents were part of the same patent family, and so the analysis is only based on the first of these patents).
As requested, this will return up to 2000 patents - in this case 1580 patents were returned. The top few patents look like this:
Three key fields in this view are:
Similarity score - the predicted similarity between the Singular Computing patent family and the listed patent families. Generally for a licensing potential analysis, we use a minimum similarity score of 1,
Licensing Potential - the predicted potential to license to these other patent owners. Licensing Potential values can be added togehter.
Owner.
Adding up the Licensing Potential values is easiest if we export these results as a spreadsheet, and then produce the following graph:
This graph ranks the different patent owners that Ambercite predicts is in this space. Intel with a total Licensing Potential of 106.6 from 43 patent families is at the top of the list,….. and right down the bottom is Google, with a total Licensing Potential of 1.1 from one patent familty.
.In fact the list extends past Google, but in this case, we may stop this list here. In terms of the aim of this blog, it is pretty clear that Google has not been filing patents similar to the Singular Computing patents family. And this would suggest they are not interested in this technology space.
So our prediction is - Google may win this case. A big call we know - so lets see how the court case plays out, and how right (or wrong) we end up being.
Has this approach been proven?
An earlier blog has provided a clear prediction of a billion-dollar litigation - you can read about it here.