I am in my 40s and yes seeing ones skills beeing Made obsolete is scary. But I still love the moment when I start with something new (language, tool, ...) whenever my job allowed me to dive into something new I am somehow grateful. And yes I do know the sentiment when I see something and catch myself thinking well this is the same as that which we had x years ago but actually it never really is or it is but now it really works. I dont fear AI will make me obsolete and for all the issues it has I am also astonished what is happening and want to know how it works. What other Industries offer this? That said I acknowledge that change comes with a lot of cognitive load which can cause Stress and anxiety and WE need ways to keep those in Check. I think this is also a Job of a company having to do with Tech, to Care for its workers (and everyone working in this industry)
Last month I needed to work in C/C++ which I usually don't, so ChatGPT helped me quite a bit, mainly to save time for researching, trial and error, etc. For example:
- I received a bunch of C files (~100) in a non-trivial folder structure from a code generator, however without a Makefile or something similar. GPT generated a Makefile based on the folder structure.
- I needed to add a couple of functions regarding reading and parsing a certain binary file structure. GPT was a good help here, too.
Additionally I was reading about some topics in linear algebra. I asked GPT to produce a intuitive explanation of the topics. This then helped me to work through the abstract description.
I recommend Andreas Reckwitz: The Society of Singularities. It changed the way I look at society. At its core is a very broad concept of status called singularities (at least that is my understanding)
In my opinion the article exagerates the claims beeing made about AI Tools ("will solve the climate crisis") just to debunk those. Are such claims really beeing made somewhere? Even the quotes inside the article are not as bold as the article claims. (e.g. Eric Schmid saying AI "can be used to support all stakeholders in taking a more informed and data-driven approach to combating carbon emissions")
In read parts of Murphys "Probabilistic Maschine Laearning" (vol 1) which is an update of an existing book in ML. It covers a broad range of topics also very recent developments. It also includes foundation topics such as probability, linear algebra, optimization. Also it is quite aligned with the Goodfellow book. I found it quite challenging at certain points. What helped a lot was to read a book on bayesian statistics.
I used Think Bayes by Allen Downey for that (http://allendowney.github.io/ThinkBayes2/index.html)
I definitely can relate to the author as well. I think we are all putting a lot of effort in standing out in some way because that is how (western?) society distributes rewards. We favor the things that are special or unique in some way. We celebrate all the possibilities that allow us be special and unique but sometimes overlook that this also generates a lot of pressure and when you fail to stand out its easy to feal inadequate.
I liked the book by Michael Sandel "The Tyranny of Merit" which goes somewhat in this direction.
I suspect that "works well" means that the model will output words in "official german" and kind of corrects pronounciation errors? I am asking because I had the use case to automatically give feedback to non native german speakers.
We had to deal with such overlapping labels a lot. For us working with separate models for each Label performed mediocre. We went for a token labeling with multiple labels per token to allow for overlaps.
From their description I also wonder wether the same span can have multiple labels?
I was reading on ropes a couple of months ago when researching emacs internals. Apparently emacs and vs code are using gap buffers. Would be interesting to know the reasons for the decisions.