I found it more weird that he specifically mentioned that the model was being developed in MATLAB, but didn’t mention any other details. To me, that sounded a little like saying “Anna and the others are writing a paper about AI risk on an Asus Eee and Google Docs”.
What I took away from the mention of MATLAB is that the model is expressed as a computer program, as opposed to just talked about, and that this requires a certain level of rigor. But yeah, I don’t care so much that it is MATLAB rather than Java.
I didn’t realize that MATLAB and Java were members of the same category. I thought that MATLAB was a software program (like Microsoft Word), while Java was a programming language (like C++).
Matlab is a program, but it’s more like the Java virtual machine than like MS Word. Both the JVM and Matlab are able to execute arbitrary programs supplied as input. So it’s meaningful to talk about Matlab-the-language. The one difference is that Java is a standardized language, whereas Matlab-the-language is “whatever Matlab-the-program accepts these days”.
MATLAB is a programming language, and I have written programs in it (though it has been a while). I would definately prefer Java (or C#) for writing most general applications, but MATLAB is good for some math heavy stuff.
This. It comes off as amateurish, not knowing which details are important to include. But hopefully these semi-informal discussions help with refining the pitch and presentation before they’re standing in front of potential donors.
Citing future publications in an academic article is a bad sign, but here Luke is telling us about SIAI’s strategy, and how what they (and related organisation FHI) have don, are currently doing, and planning to do fits into that that strategy. Discussing current research projects in this case is good.
Referencing future publications/results detracts from one’s credibility.
I found it more weird that he specifically mentioned that the model was being developed in MATLAB, but didn’t mention any other details. To me, that sounded a little like saying “Anna and the others are writing a paper about AI risk on an Asus Eee and Google Docs”.
What I took away from the mention of MATLAB is that the model is expressed as a computer program, as opposed to just talked about, and that this requires a certain level of rigor. But yeah, I don’t care so much that it is MATLAB rather than Java.
I didn’t realize that MATLAB and Java were members of the same category. I thought that MATLAB was a software program (like Microsoft Word), while Java was a programming language (like C++).
Matlab is a program, but it’s more like the Java virtual machine than like MS Word. Both the JVM and Matlab are able to execute arbitrary programs supplied as input. So it’s meaningful to talk about Matlab-the-language. The one difference is that Java is a standardized language, whereas Matlab-the-language is “whatever Matlab-the-program accepts these days”.
MATLAB is a programming language, and I have written programs in it (though it has been a while). I would definately prefer Java (or C#) for writing most general applications, but MATLAB is good for some math heavy stuff.
This. It comes off as amateurish, not knowing which details are important to include. But hopefully these semi-informal discussions help with refining the pitch and presentation before they’re standing in front of potential donors.
Yup. Many cranks say things like “I programmed a human-level AI in Fortran!”, so it’s a bad pattern match to enable.
Citing future publications in an academic article is a bad sign, but here Luke is telling us about SIAI’s strategy, and how what they (and related organisation FHI) have don, are currently doing, and planning to do fits into that that strategy. Discussing current research projects in this case is good.
And Luke has not cited any future results. That would be crazy.