Two events next weekend

[dmm]

An online talk on efforts towards practical use of AI for assistance in software engineering at Google (this is California time):

www.meetup.com/Scala-Bay/events/271129752/

("Towards an ML-augmented Programming Stack" by Eugene Kirpichov. A slide deck is attached to the event page, the key slides are slides 11-13.)

An online demoparty (this is Boston time):

atparty-demoscene.net/

One can just watch, or one can take part in various competitions.

Comments

[dmm]

> we are not yet close to a computer system which would be able to read a textbook and learn to reason about algorithmic complexity from that

is that correct, actually?

Look at this paper from Allen Institute for AI:

https://arxiv.org/abs/2002.05867

"Transformers as Soft Reasoners over Language"

'This paper investigates a modern approach to this problem where the facts and rules are provided as natural language sentences, thus bypassing a formal representation. We train transformers to reason (or emulate reasoning) over these sentences using synthetically generated data. Our models, that we call RuleTakers, provide the first empirical demonstration that this kind of soft reasoning over language is learnable, can achieve high (99%) accuracy, and generalizes to test data requiring substantially deeper chaining than seen during training (95%+ scores). We also demonstrate that the models transfer well to two hand-authored rulebases, and to rulebases paraphrased into more natural language. These findings are significant as it suggests a new role for transformers, namely as limited "soft theorem provers" operating over explicit theories in language.'

perhaps, we might be closer than it seems...