The societal impact of artificial intelligence – What artificial intelligence can do in the legal system and how
Anna Ronkainen Chief Scientist, TrademarkNow @ronkaine AIHelsinki kickoff 2015-12-16
$200–300B USD/year!
A lot of difficult conversations to be had about AI and society - will full employment be just a pipe dream – and
is it even desirable anymore - will income equality increase even more if
people are divided into those who tell computers what to do and those who do what computers tell them to do
- will we still allow people to drive cars - will big data and omnipresent data gathering
make privacy a thing of the past - ...oh, and about those autonomous weapons...
Presuming we’re a more solution-oriented bunch in this room... - commercially AI is now where mobile was in
the ~1980s: the only way is up - probably no new N-word in sight, instead
many smaller Finnish companies doing quite well (look around you)
- AI potentially affecting 100–200k jobs in Finland alone by 2025, and that tech has to come from somewhere
What’s law got to do with it? - one of the very first application domains for AI
(1950s) - back in the day, law was at the forefront of the
automatic data processing revolution in general (e.g. the great Pennsylvania Health Code search-and-replace)
- kind of a structural isomorphism between law (as commands from the legislator to be carried out by a judge or a citizen) and software (commands from the programmer to be carried out by the CPU), superficially correct but misleading
In ye olden days (and still in Finland) there were legal informatics - everything having to do with computers and
the law lumped into one discipline - software copyright, patents, privacy... - computational legal theory - theory of legal information
- originally a 50/50 mix of people with a (often practical) legal background and a CS background, until the 1990s
More and less recent trends in AI & law - 1980s: expert systems (and logics) - 1990s: ontologies (and logics) - 2000s: argumentation (and logics) - 2010s: outside interest/wake-up call from e-
discovery (and logics)
- a research community of ~500 people - main confs ICAIL and JURIX; AI&law journal
published by $pringer
A couple of real-life examples
(Ronkainen (2010): Mosong, a Fuzzy Logic Model of Trade Mark Similarity)
What we do at TrademarkNow - trademark search: making sure your new
brand isn’t too close to earlier trademarks to give your problems
- trademark watch: alerting you about new filings too close to your own marks to help you take the necessary steps to protect them (by filing an opposition against the new mark)
- globally, now >60 jurisdictions fully covered
How we do it - data acquisition, import - inbound processing of individual marks, storage - search and watch UIs - likelihood of confusion algorithm - similarity of trademarks
- phonetical, graphical, semantic, animal, mineral - similarity of goods and services
- registrability (absolute grounds) analysis and other useful information (e.g. dictionary results)
- reporting - ...all using all kinds of AI techniques from GOFAI to
deep learning as appropriate
Why Finland needs more AI & law - fewest lawyers per capita in the OECD - court system struggles esp. with processing
times (many, many ECHR judgements) - most solutions have to be jurisdiction-
specific (because the law is so different) - small country = small market, not all that
interesting for outsiders
Why AI & law needs Finland - the AI & law research community has been
rather insular (focus on just a couple of special topics at a time, no new ideas coming from the outside)
- little focus on building systems with a practical impact (or even validation, or even doing work that can be validated)
- ...and (of course) lots of amazeballs AI people with mad skillz in Finland
Thank you!