Monthly Archives: August 2018

Google Play Music Manager says “MSVCP120.dll was not found”

After an update to Windows 10, Google Play Music Manager started giving this error on startup:

“The code execution cannot proceed because MSVCP120.dll was not found. Reinstalling the program may fix this problem”

Well, reinstalling didn’t fix it. The dll mentioned does exist in the Windows system32 folder, but it turns out that it is the 64 bit version, and the 32 bit version is needed for Music Manager to work (you can have both installed side by side).

Solution: go to https://support.microsoft.com/en-us/help/3179560/update-for-visual-c-2013-and-visual-c-redistributable-package and scroll down to the x86.exe section. Install the correct language for you. Here is a direct link to the English one for quickness: http://download.microsoft.com/download/0/5/6/056dcda9-d667-4e27-8001-8a0c6971d6b1/vcredist_x86.exe

After installing the Visual C++ 2013 Redistributable (x86) package, it works again:

BBC 4.1 AI TV

Intriguing advert on BBC Four suggesting a two night experiment into AI.

I could only find this article about it, which suggests artificial intelligence and machine learning could be used to choose programmes from the BBC archive to schedule a couple of evenings worth of shows on BBC Four. Which sounds a bit underwhelming. Hopefully it turns out to be something more impressive.

If anyone knows more, drop something in the comments, I would be interested to hear about it!

Update: looks like someone has added the advert to YouTube.

Update 29/8/18. A new BBC page has appeared about this: https://www.bbc.co.uk/mediacentre/latestnews/2018/bbc-four-ai

So, on 4th and 5th of September, there will be programmes on about AI and robots. The experimental part is the programme “Made By Machine: When AI Met The Archive” on Wednesday 5th at 9pm. It will have 4 sections. The first 3 will each use a different method to search the BBC archives in order to find content clips that it thinks it can stitch together based on similarity, first based on visual content, then based on language, and then based on the type of activity on screen (high or low energy scenes). The 4th section will use all three methods to create a new piece of content from the archive. So it could be quite interesting actually. I am interested in whether it is going to be doing this in real or near-real time, as the amount of content to search through must be enormous.

What is the point? It seems one aspiration is to be able to generate related content based on a topic. E.g. think of a breaking news story. If the AI was able to pull out things from the past about the person, place or thing in the news, and put it together into a programme automatically, it could give the BBC the edge over other outlets who are relying on humans trawling through information manually.