Agents and grammars
Making more progress towards the end of term, doing a little reading regarding the agency of AI, and generally ticking over. It’s been a good week. I’ve ticked off another major coursework that had to be done before I finish for the semester. Aside from that everything else is making nice progress.
I haven’t really left the house since the beginning of this second lockdown. I’ve been out for walks and exercise, and once or twice to the shops, but apart from that I’ve just been sticking it out with work and my view of the pigeons. This is probably the number one reason everything is going so well for me as a graduate student. A friend pointed this out to me on a call today, I’ve always been a bit of a nerd, but I also love life. Unlike during my undergrad, there’s not an awful lot of life to go around loving at the moment, so I’m just nerding out every day.
Parsing
This week part of the play has been with parsing. We’ve been asked to implement a brute force parser that takes a context-free grammar, and a word, and returns the parse tree of that word if it is in the grammar. I initially completed this with the CYK algorithm before I found out the choice was a little more strict. The CYK algorithm has the benefit of $O(n^3 \times |G|)$ complexity, which seems high but is still a lot better than brute force, where you generate all possible derivations for step length of $2n - 1$, where n is the length of the word in question. I’m not sure what the learning goal is in having us use the brute force algorithm, but I have learned how quickly that complexity balloons. I haven’t yet played around with probabilistic parsing which seeks to reduce the best case complexity, but that sounds interesting.
Robot agency
On another side I’ve been introduced to the writing of Joanna Bryson, who was at Bath until very recently. Ethical AI, and the ethics of robotics/technology seem like absolutely fascinating areas, so I’m finding opportunities to bend my work towards that. At the moment I’m just wriggling my way through some of the most central papers in the area, and hopefully I’ll be able to develop a better sense of what’s going on. This is something I need to read more on, but questions of whether we would ever want to grant robots or AI agency, and who would that serve are really engaging. Picking up Patiency is not a virtue and Can a Robot Be a Good Colleague? to get started.
I feel that early struggle of trying to get the lay of the land in a discipline at the moment. It’s still a struggle since I’m still discovering what I don’t know enough about, the unknown unknowns for me. I’m only just going through the foundations with Turing Machines so getting to a place where I feel I can get the conversations happening feels a bit overwhelming, more so than it did when I was tackling an English degree (somehow the stakes for originality felt lower there, not to insult it at all). Anyway, it’s all forward progress.
Brief rant
I’m frustrated again by the false dichotomy presented repeatedly in the UK of trading lessened economic impact for lost lives during the COVID-19 response and with lockdowns. Good public health has massive economic benefits. Upstream interventions save lives and protect economic upsides. Similarly lack of upstream public health interventions negatively impacts the economy.
Following on from this article in the FT today the UK has now spent more as a proportion of GDP than any other G7 country to mitigate the pandemic, and it has had the worst outcome both in health and economic terms. The research points to our late lockdowns, and dithering decisions. Delaying the lockdowns (both the first and the second time) beyond the point at which it was clear they were needed means they have to go on for much longer, and have lesser impact. The leadership of the country has failed in so many ways, but there seems to be no ongoing acknowledgement or response for the failures. It feels like criticism of this government, which has now been in power since 2010 in some form and iteration of leadership, hasn’t been punching through. It’s an extremely depressing way to see your country go as public trust is hammered through inept and corrupt governance. The only thing that seems to be discouraging young mobile adults from emigrating is the fact that from the 1st of January we can no longer move and work freely in the Schengen Area. There’s light at the end of the tunnel with the vaccines coming at least.
—