Errant thought from June 13:
The world keeps seeming more and more dystopian. But it is really amazing to be able to ask a computer complex things in natural language, and get back coherent, valid responses in natural language too. My mind often glosses over that wonder because of everything else going on. And also because of knowing that wonderful achievement has its own associated dystopian aspects to it, which will just get worse and worse as they are exploited.

.

Like this: The AI Slop Fight Between Iran and Israel

.

Kids growing up now and in the future may find it difficult to believe actual photos and movie footage from past atrocities, including the Holocaust. They will have no way to trust that the images are real and were not generated by AI. Perhaps if they looked through an old book with photos, and could trust that it was published in the year it said it was, before AI was capable of what it is now, perhaps that would be believable to them. But books are becoming rarer, especially old ones. Old books didn't have many photos. Were there many books with photos of atrocities? Encyclopedias would have perhaps a few photos for each topic. Certain magazines were more likely to have extensive photos. There may be digitized versions of the books and magazines available online, but then you get back to how can you trust that the digitized item you're looking at isn't a fake?

I'm already at that distrusting stage with much of what I see online. When you don't know what is trustworthy, you end up choosing to believe the things which align with your already-formed beliefs. "This aligns with my beliefs; I don't know if it is true, but I wouldn't be surprised if it was." "This doesn't align with my beliefs; why spend time and mental effort considering that it might be true when it very well might not be?"

While shopping for bed slats (not for me), I found a set on the Kohl's website that says it is made "right here" in the USA from Canadian wood. But while checking out that brand's website, Continental Sleep, I belatedly realized the wording and grammar on their pages isn't quite right. So now I wonder, is the "Made in USA" claim a lie? Or do Chinese companies have factories in the U.S. making that kind of stuff? I could have sworn I saw that "Made in the USA" claim on one of their website pages too (for only that one product, which made it seem all the more dubious), but am not finding it anymore.

I looked up info on the company but found nothing other than a reference to "Continental Sleep Holdings", which was likely a different company, on a Wikipedia page about "Sleepeezee".

The reviews on the ContinentalSleep website seem fake too, with similar grammar and wording issues; not the kind Americans would make.

It's a bit surprising they haven't used an LLM to fix the grammar on their webpages. Once Chinese companies get better at doing that, there won't even be that tell to give them away.

I came across this song on Youtube: The Cover Girls - Show Me Hearthrob Mix + Drumapella

The video shows an image of the vinyl record listing two separate songs on it, "Show Me (Heartthrob Mix)" and "Drumapella". I was curious what the 2nd song sounded like and wasn't clear if the video included it or not.

So I searched on that song title, "Drumapella", and was confused that the results listed many other songs by other artists with that word in their title. Searching on "what is drumapella" didn't give an answer. Wiktionary doesn't have the word listed either.

So I asked Copilot. This was the response:
"Drumapella" typically refers to a musical track or piece that combines the elements of a drumline (or drum track) with a cappella singing, which means singing without instrumental accompaniment. It's essentially a fusion of rhythmic drumming and vocal harmonies.
It’s an innovative way to blend percussive beats with vocal performances, creating a unique and dynamic sound.

Oh! So, "drum" + "a capella" = "drumapella". It makes sense.

Then I asked Copilot "What are some webpages that explain that?" The reply:
I couldn't find specific webpages that explain "drumapella" in detail, but there are some YouTube videos that showcase drumapella performances. These videos might give you a better understanding of how drumapella works...

It then gave links to 3 videos of songs with "drumapella" in the titles, none of whose descriptions explain what the word means. All 3 songs are from the same YouTube account, DjRubyEve. That account has been posting several remixes of famous songs per day, so I suspect the account uses AI to generate the remixes.

Other Youtube accounts have also posted "drumapella" songs, however. Some of these aren't drums + acapella singing, but rather only drums - a different usage/meaning of the word, which also makes sense.

A 1998 album with remixes of Madonna's song "Frozen" also has two "Drumapella" versions, so it is not a new word.

I am impressed that an LLM was able to figure out (presumably on its own, since I didn't find the word defined anywhere) that drumapella might be a combination of drums and acapella, especially as that didn't even occur to me until the LLM mentioned it.

.

I have been turning to the LLMs for answers more and more often when my web searches prove unilluminating.

.

When I originally searched on "what is drumapella", this Reddit page was one of the results: Is there a musical term for "a capella but percussion is allowed"?
I glossed over that page, because when I searched it, "drumapella" was not found. But now that I expanded all the collapsed comments in the thread, I found this one reply:
Dub?
”Drumcapella”?

So maybe the LLM partially got the answer from there even though "Drumcapella" has an extra "C" in it compared to "Drumapella".

I saved some screenshots to show how easy it can be (in some cases) to get Copilot to help you with simple coding tasks.

These were my prompts:
1. Hello, please suggest how to code a powershell script to convert values like "1724143246792" to a timestamp format of YYYYMMDD-hhmmss

2. The script should take the input value as a parameter

3. Thank you but it isn't giving the right date. 1724143246792 should convert to approximately 20240819-132248, but the above script gives 16010102-235334

4. Thank you, that works better. But the output time is in UTC and I would like it in Eastern US time

5. Thank you, that is good. If the input value is from the summer, will the output be in EDT, or will it always be in EST?

( Screenshots of Copilot's answers behind the cut )

Previously when I tried out dark mode for various applications on my laptop, it made text seem less coherent to me, so I didn't use it.

But recently the normal (light) mode started bothering my eyes (they've been getting very tired & dry & achy), so I switched both my work and personal laptops into dark mode in the Windows settings. I've been switching individual applications into dark mode too, where possible.

My spreadsheets have been giving me difficulty. They are LibreOffice Calc ODS files. For whatever reason, most of the text in them is set to black, not "automatic". So when I switch to the "LibreOffice Dark" color scheme, I end up with black text on a black background, impossible to read.

I have many spreadsheet files, with many tabs (sheets) in each file.
There ought to be an easy way, a script, to update multiple spreadsheet files at once to convert all text, current and future, to the "Automatic" color. But I haven't found such a script yet.

Doing it manually involves going it each tab in each file, selecting all the cells, and updating the text color. And then doing something else to change the default for new cells too. I did the first step for two spreadsheets. Then I noticed that text I enter in a row below the last existing row ends up black on black again. That's why I think the 2nd step is required too (and I don't even know how to do that step manually yet).

I asked Gpt4o-mini (via DuckDuckGo) for assistance. It suggested creating a macro, and gave me sample code for the macro. The sample code was giving me an error in the part for finding all ODS files in a specified folder. I worked through that error with Gpt4o-mini's assistance. Now the part for updating each file is giving me an error (probably the first of several). Sigh. I don't want to spend my vacation time trying to code and debug a spreadsheet macro. Debugging the macro isn't as easy as debugging Java code in Eclipse. You can't just select a variable in the code and see all the details for it.

Surely such a macro or script already exists somewhere. I can't be the first person facing this problem.

I get stuck in all these problems. How can i make progress. How can I ever get to sleep at a good time. How can I ever catch up.

UPDATED:
I tried asking Copilot. There were a few issues with the first code it suggested too, but we fixed them. This logic seems to work well to update the text color to "Automatic" in all ODS files in the specified folder:

Sub ChangeTextColorInFolder()
    Dim oDoc As Object
    Dim oSheet As Object
    Dim oCell As Object
    Dim oRange As Object
    Dim oCursor As Object
    Dim sFolderPath As String
    Dim sFileName As String
    Dim i As Integer
    Dim oSimpleFileAccess As Object
    Dim aFiles() As String
    Dim args(0) As New com.sun.star.beans.PropertyValue

    ' Set the folder path
    sFolderPath = "C:\path\to\your\folder\"

    ' Get the SimpleFileAccess service
    oSimpleFileAccess = CreateUnoService("com.sun.star.ucb.SimpleFileAccess")

    ' Get the list of files in the folder
    aFiles = oSimpleFileAccess.getFolderContents(ConvertToURL(sFolderPath), False)

    ' Loop through all the files in the folder
    For i = LBound(aFiles) To UBound(aFiles)
        sFileName = aFiles(i)
        
        ' Check if the file is an ODS file
        If Right(sFileName, 4) = ".ods" Then
            ' Open the ODS file in hidden mode
            args(0).Name = "Hidden"
            args(0).Value = True
            oDoc = StarDesktop.loadComponentFromURL(sFileName, "_blank", 0, args)

            ' Loop through all the sheets in the document
            For Each oSheet In oDoc.Sheets
                ' Get the used range of the sheet
                oCursor = oSheet.createCursor()
                oCursor.gotoEndOfUsedArea(False)
                oRange = oSheet.getCellRangeByPosition(0, 0, oCursor.RangeAddress.EndColumn, oCursor.RangeAddress.EndRow)

                ' Loop through all the cells in the range
                For j = 0 To oRange.Rows.Count - 1
                    For k = 0 To oRange.Columns.Count - 1
                        oCell = oRange.getCellByPosition(k, j)
                        ' Set the text color to Automatic
                        oCell.CharColor = -1
                    Next k
                Next j
            Next oSheet

            ' Modify the default cell style to automatic color
            Dim oStyle As Object
            oStyle = oDoc.StyleFamilies.getByName("CellStyles").getByName("Default")
            oStyle.CharColor = -1

            ' Save and close the document
            oDoc.store()
            oDoc.close(True)
        End If
    Next i

    ' Show a message box to indicate completion
    MsgBox "Macro completed successfully!", 64, "Success"
End Sub

That is for both the first and 2nd steps.
For how to do the 2nd part manually, this answer works:

To ensure that all cells, including those outside the used range, default to Automatic color, you can modify the default cell style in LibreOffice Calc. Here’s how you can do it manually:

Open LibreOffice Calc.
Go to Format > Styles and Formatting (or press F11).
In the Styles and Formatting window, click on the Cell Styles icon (second icon from the left).
Right-click on Default and select Modify.
In the Cell Style dialog, go to the Font Effects tab.
Set the Font Color to Automatic.
Click OK to apply the changes.

I'm finding it hard not to like these AI-generated videos. Even though I think it is wrong and probably illegal to use the actors' likenesses without permission and compensation. Or at least... I find the videos quite interesting, and the redneck twist is different. It is hard for me to tell which clips of people (if any) are real recordings and which are generated.

As for the Harry Potter one, I don't want to watch or share anything that will benefit the author. But after watching several of these videos, I doubt the entities who made these videos got anyone's official permission to do so. So I also doubt any royalties are getting paid. Correct me if I'm wrong? Maybe YouTube somehow funnels money to the correct people?

Harry Potter - Redneck Wizard 2 (Official Music Video)
https://www.youtube.com/watch?v=NG4XeBOVBw4

Star Wars - Revenge of the Redneck (Official Music Video)
https://www.youtube.com/watch?v=GtvlCyJNWPE

Matrix - Redneck Reloaded (Official Music Video)
https://www.youtube.com/watch?v=7CJTWOoS8Qs

I watched a course on Generative AI Prompting today where one of the hands-on tasks was to get ChatGPT to generate rhyming questions and answers from a paragraph taken from Wikipedia.

I took this paragraph on Saturn:

Saturn's interior is thought to be composed of a rocky core, surrounded by a deep layer of metallic hydrogen, an intermediate layer of liquid hydrogen and liquid helium, and finally, a gaseous outer layer. Saturn has a pale yellow hue due to ammonia crystals in its upper atmosphere. An electrical current within the metallic hydrogen layer is thought to give rise to Saturn's planetary magnetic field, which is weaker than Earth's, but which has a magnetic moment 580 times that of Earth due to Saturn's larger size. Saturn's magnetic field strength is around one-twentieth of Jupiter's. The outer atmosphere is generally bland and lacking in contrast, although long-lived features can appear. Wind speeds on Saturn can reach 1,800 kilometres per hour (1,100 miles per hour).

I couldn't think of any rhymes from that myself, to provide as "few-shot" examples. So I simply asked ChatGPT to generate rhyming Q&As from the text.

It did pretty good; I liked these rhymes the best:

Q: Why does Saturn have a yellowish glow?
A: Ammonia crystals in clouds, you know,
Give it a pale hue, a soft show,
In the upper atmosphere, where they grow.

Q: What's the look of Saturn's outer layer?
A: Bland and lacking contrast there,
Yet long-lived features can appear,
In the atmosphere's drifting air.

I browsed through a Goodwill store. The music CDs were the first thing to catch my eye; I looked through some of them. I don't really want CDs as that requires me ripping the music to MP3 files, and leaves me with a physical CD left over taking up space. But it reminds me of the old days, looking at CDs in stores and taking a chance on some, buying them, never knowing for sure how much I was going to like the music until I got home to listen to it.

One of the CDs that caught my eye was in a foreign language. It looked like Finnish to me, but I wasn't sure. The name on it was Yva plus a last name that I don't remember. It may have started with B. The back side had a picture of someone (presumably Yva the singer) in a blue and white outfit, similar to a traditional Bavarian style dirndl.
Another CD had music from RiverDance.

I didn't end up getting any CDs. I browsed the rest of the store, intending to possibly stop by the CD rack again afterward if I hadn't talked myself out of it by then. But then as I walked by the registers in the direction of both the entrance and the CDs (I'm not sure anymore which I was headed for), the lone cashier there said 'goodbye' or 'have a nice day' or something like that to me. Then when I thought of the CDs, it seemed awkward to stay and look at them again when {goodbyes had already been made}. I stopped in the sunshine of the open doorway and checked my phone messages, mentally debating it. Decided to leave.

At home I looked up this singer Yva. I only found an English singer and a Lithuanian group by that name.
Maybe it was this album by the Lithuanian group Yva, but I don't think so. I think the CD I looked at had only 1 person on it, not 3. And I remember there being a last name along with the Yva. It seemed like an older album. I don't remember what was on the front of the CD. Oh well. I probably wouldn't have liked the music anyway.

..

I came across this name website:
https://www.nameuc.com/talon-name-meaning-origin-popularity/

It has a different pretty image for each name. They are fairly obviously generated by an AI/LLM/whatever it is called for images. A while later I realized the text of the site must have also been generated by a LLM too. (It sounds quite similar to the answers I got when I asked an LLM questions about names.) That indicates to me that the data listed for the names on that site is not very trustworthy. If it hadn't been for the novel pretty name pictures, I might not have noticed/realized that. Many of the other name sites I've come across this last year or more have been similar, I suppose, just without the images. Sigh. How are we going to filter the junk data from the accurate data?

..

Apparently the aurora was visible even near here in South Carolina on Friday night. I didn't read/hear about it until this morning when the sun was already starting to rise. I never imagined an aurora could/would be visible this far south. I remember seeing the crescent moon in the sky earlier while I was out shopping. The sky was normal, not ablaze with colors.

Beautiful mural painted on a new local vegan restaurant:
part 1: https://www.instagram.com/morevegancafe/reel/C4k4l8sOX2X/
part 2: https://www.instagram.com/morevegancafe/reel/C40k1wruAUf/
The artist: https://www.instagram.com/ija_monet/

Neat AI-generated video of morphing martial arts anime characters:
https://www.instagram.com/djvoodoochild/reel/C2wgWy1uQMx/

Nice dance moves:
https://www.instagram.com/djvoodoochild/reel/CxlYgtCPwGv/

ChatGPT achieves the pinnacle of human intelligence, laziness, and developers are baffled (2023/12/12)

This AI can pick up passwords from the sound of your keystrokes (2023/12/10)
This is something I've been slightly concerned about for some time already. From that article, it doesn't sound very advanced yet - the AI needs to be trained on each specific keyboard's sounds first. But I am sure they will get better at it over time, so as not to need that initial training in the future. And I bet some government security agencies have more advanced versions like that already.

Utah Supreme Court says suspects can refuse to hand over phone passwords to the police. Other state Supreme Courts disagree and the case would wind up before the US Supreme Court (2023/12/16)
The state Supreme Court also noted that the case raises important questions about how the Fifth Amendment extends to law enforcement efforts to unlock smartphones. The justices noted, as an example, law enforcement obtaining an order to compel a suspect to provide an unlocked device, thus circumventing the necessity of having them disclose the password.

With the Valdez case, the police asked him to verbally provide his password and did not get an order to compel him to unlock the device. ...