| transcript | |
|---|---|
|
00:00 - 0:00:02
*Music*
|
|
|
0:00:02 – 0:00:31
[Music]
|
|
|
0:00:31 – 0:00:39
Hello and welcome to a new issue of Abspalfung, where I occasionally actually
|
|
|
0:00:39 – 0:00:47
I almost never skimp on the topic, unless I actually skip it. I'm Sascha
|
|
|
0:00:47 – 0:00:56
Machman and say hello. Today's topic is why I get five euros from the federal government.
|
|
|
0:00:56 – 0:01:06
I got it. You don't go to the mailbox and look for the mail and then
|
|
|
0:01:06 – 0:01:15
You can find a letter from an institution affiliated with the federal government.
|
|
|
0:01:15 – 0:01:25
You open the thing and find a letter inside that you've been selected by a
|
|
|
0:01:25 – 0:01:35
A whole bunch of people in my neighborhood and as a little thank you you you would have a little
|
|
|
0:01:35 – 0:01:43
Gift packed with it and there was a small envelope in it and there was quasi on it
|
|
|
0:01:43 – 0:01:52
"Thank you" and such a gift, a picogram of a gift, today you would say emoji
|
|
|
0:01:52 – 0:01:59
of a gift. I opened this up and there were really old ones,
|
|
|
0:01:59 – 0:02:09
5 euros in it. A really broken, shabby 5 Euro bill, where I thought,
|
|
|
0:02:09 – 0:02:18
The government has the money, so why don't they issue new bills? But only on the sidelines.
|
|
|
0:02:18 – 0:02:24
On the letter was a QR code with which you came to the survey.
|
|
|
0:02:24 – 0:02:27
Otherwise there was just a little bla bla bla bla in there.
|
|
|
0:02:27 – 0:02:31
And I found that basically interesting.
|
|
|
0:02:31 – 0:02:34
I often do surveys like this.
|
|
|
0:02:34 – 0:02:36
I did this for Infratest.
|
|
|
0:02:36 – 0:02:40
Interestingly enough, I've been selected time and time again.
|
|
|
0:02:40 – 0:02:44
to participate in such surveys.
|
|
|
0:02:44 – 0:02:53
So this is sometimes about these typical standard questions, what is the highest level of education,
|
|
|
0:02:53 – 0:03:02
you have, how many family members and such a thing has the household, how many people
|
|
|
0:03:02 – 0:03:12
Living in the household and this time the survey was a bit different. It was about learning
|
|
|
0:03:12 – 0:03:21
for life almost as far as one has used educational offers to continue education. So it
|
|
|
0:03:21 – 0:03:27
was not so typical of these data collections, what is the annual income of the household and so
|
|
|
0:03:27 – 0:03:33
What, but it was quite specific. I thought it was cool and very interesting.
|
|
|
0:03:33 – 0:03:41
Yes, and you could actually finish the story now, because I need
|
|
|
0:03:41 – 0:03:47
Now, don't just list what I've answered. It is also
|
|
|
0:03:47 – 0:03:55
boring to death. But so far I've never heard of anyone for a
|
|
|
0:03:55 – 0:04:01
The survey received five euros as a small thank you. I think this is
|
|
|
0:04:01 – 0:04:09
Something I'll frame myself for. I'm going to hang it on the wall because I
|
|
|
0:04:09 – 0:04:17
I find this so strange, the fact that this is being sent.
|
|
|
0:04:17 – 0:04:21
And then you ask yourself, how expensive is that?
|
|
|
0:04:21 – 0:04:25
What is the cost of such things?
|
|
|
0:04:25 – 0:04:32
So I talked to my acquaintances and colleagues about it and nobody knows
|
|
|
0:04:32 – 0:04:35
Someone who has experienced this for themselves.
|
|
|
0:04:35 – 0:04:45
Like I said, I do this more often. Sometimes I get annoyed by the questions,
|
|
|
0:04:45 – 0:04:57
Then I get annoyed every time, what did you get into and why do you even do that?
|
|
|
0:04:57 – 0:05:07
Because these questions are repeated, but then they are just changed. So it's almost
|
|
|
0:05:07 – 0:05:12
The one. Once the question is asked in such a way that you can answer it with yes and the next
|
|
|
0:05:12 – 0:05:17
Sometimes it is set in such a way that you can answer it with no. I know why you do this,
|
|
|
0:05:17 – 0:05:26
This is almost double verification. This means that it is impossible for anyone to
|
|
|
0:05:26 – 0:05:33
Just click through so randomly, because that is then just conspicuous in the statistics. I mean,
|
|
|
0:05:33 – 0:05:39
I did this in my surveys, which I did as part of my studies, even so.
|
|
|
0:05:39 – 0:05:46
made. And even then again after a long time, so if there are now several questions in between
|
|
|
0:05:46 – 0:05:52
I have always asked such verification questions. This is a total
|
|
|
0:05:52 – 0:06:00
"Legitimate means". Only if you see through this, then it really annoys,
|
|
|
0:06:00 – 0:06:08
Then it annoys me hard. I've really done a bunch of these surveys in my life
|
|
|
0:06:08 – 0:06:15
Interestingly, the most recent surveys were mostly scientific.
|
|
|
0:06:15 – 0:06:24
Interviews in the context of the podcast world. I'm on different forums and I think
|
|
|
0:06:24 – 0:06:31
one of the most important is the Sende-Gate and there are regularly young researchers
|
|
|
0:06:31 – 0:06:40
Just ask me what it looks like right now and so on. And for them it is interesting that they
|
|
|
0:06:40 – 0:06:49
Not only from a listening point of view, but also as a maker. How are people,
|
|
|
0:06:49 – 0:06:54
Who deals with podcast production, is prepared for certain complex questions?
|
|
|
0:06:54 – 0:07:03
Yes, and this is always interesting when you have made such questionnaires yourself,
|
|
|
0:07:03 – 0:07:11
Look at how they are done now. And sometimes I really ask myself the question,
|
|
|
0:07:11 – 0:07:17
What is the scientific value? Because if you are in such super small groups such
|
|
|
0:07:17 – 0:07:26
Surveys make, yes, that's sometimes such a distortion. Well, but they'll be
|
|
|
0:07:26 – 0:07:32
You know what they're doing. I am always happy to be informed about the results,
|
|
|
0:07:32 – 0:07:42
Because I always find it exciting when I've done it. The worst I found
|
|
|
0:07:42 – 0:07:50
so the surveys, so these basic surveys of Infratest and what they are all called, if you
|
|
|
0:07:50 – 0:07:58
This is supposed to create the basic mission, which is what the next surveys will result in.
|
|
|
0:07:58 – 0:08:04
I always find them annoying, because they are always extremely extensive questionnaires.
|
|
|
0:08:04 – 0:08:07
You should also be very careful where you do it.
|
|
|
0:08:07 – 0:08:19
Because one should only, yes, what is the name of the word now, take survey institutes,
|
|
|
0:08:19 – 0:08:25
They are also, as they say, classified, verified.
|
|
|
0:08:25 – 0:08:34
So preferably stop those with the ARD or CTF and as they are all called, Infratest for example.
|
|
|
0:08:34 – 0:08:41
With that, in which you can make these surveys, definitely. But there are also some,
|
|
|
0:08:41 – 0:08:45
They just grab the data and then you get quite a lot of advertising mail.
|
|
|
0:08:45 – 0:08:51
Interesting for things you're interested in. No wonder,
|
|
|
0:08:51 – 0:08:59
They were given the data for it. Speaking of data. I've spent the last few days
|
|
|
0:08:59 – 0:09:08
very much to program and have found a nice way to
|
|
|
0:09:08 – 0:09:15
Context-based searches. We all know how to look for keywords. When I go to
|
|
|
0:09:15 – 0:09:25
Example now, what could you be looking for now? Frank Battermann. Then you type it in and then
|
|
|
0:09:25 – 0:09:32
All text passages are found where the word Frank Battermann occurs, i.e. the two words.
|
|
|
0:09:32 – 0:09:36
If you make it a little bit smarter, then you could say,
|
|
|
0:09:36 – 0:09:41
that one works with bolic operators and then all positions would be selected,
|
|
|
0:09:41 – 0:09:46
where Frank is present, all places where Frank, Battermann is present, all places where Battermann is present
|
|
|
0:09:46 – 0:09:55
All places where Battermann and Frank occur. That's stupid. It is better,
|
|
|
0:09:55 – 0:10:00
If you have context-based search. For example, if you're looking for a job like,
|
|
|
0:10:00 – 0:10:10
There the junk radios or junkcasters met with others. That would be interesting,
|
|
|
0:10:10 – 0:10:18
if you have different possibilities, different, i.e. different meaningful
|
|
|
0:10:18 – 0:10:26
Possibilities like these junk radios again everywhere, everywhere you can find junk radio,
|
|
|
0:10:26 – 0:10:37
Scrapcast and stuff like that. But how do you do that? Put each word in a vector.
|
|
|
0:10:37 – 0:10:46
What is a vector? These are directions, simply broken down. You could still do that now
|
|
|
0:10:46 – 0:10:53
Explain it much more complicated. But if I now have a vector that indicates a direction,
|
|
|
0:10:53 – 0:11:01
there are usually three values, i.e. x, y and z. If I go now, for such a
|
|
|
0:11:01 – 0:11:09
context-based search these vectors instead, then this is in a simplest case has the
|
|
|
0:11:09 – 0:11:19
1521 dimensions. It's a bit more like three. So we move in one
|
|
|
0:11:19 – 0:11:27
Three-dimensional space. Depth, width, height. If I now add a fourth dimension,
|
|
|
0:11:27 – 0:11:34
Let's say time, then I could also observe this width, height, depth over time.
|
|
|
0:11:34 – 0:11:44
And see a change in that. Now, however, for this context-based search, each vector has 1500
|
|
|
0:11:44 – 0:11:53
blablablabla dimensions. This makes the amount of data quite high because these vectors, these individual
|
|
|
0:11:53 – 0:12:05
Dimensions are described, I believe, in 10- or 12-digit numbers. There will be of course
|
|
|
0:12:05 – 0:12:13
From a 10-kilobyte-sized message likes times 1.5 megabytes. But the benefit that you
|
|
|
0:12:13 – 0:12:20
As you can see, it is gold. I could, for example, if I used this data in ChatGBT
|
|
|
0:12:20 – 0:12:24
Use the language model to analyze it.
|
|
|
0:12:24 – 0:12:29
ChatGBT ask where did Frank Battermann say, for example,
|
|
|
0:12:29 – 0:12:39
"The junk box is stupid." For example. Purely hypothetical.
|
|
|
0:12:39 – 0:13:02
Then would be looked at, not only for "the scrap caste is stupid", but for all sorts of synonymous, meaningful similarities.
|
|
|
0:13:02 – 0:13:12
Instead of "stupid" would then be "stupid" or "non-competent", "no cognitive transfer performance"
|
|
|
0:13:12 – 0:13:14
For example, they are being sought.
|
|
|
0:13:14 – 0:13:18
And then there would be results that have a scoring.
|
|
|
0:13:18 – 0:13:25
So an assessment of how close this meaning is to the original question.
|
|
|
0:13:25 – 0:13:30
And then the results are issued almost like such a ranking.
|
|
|
0:13:30 – 0:13:38
Of course, these can be various places in such a text document that are closer or
|
|
|
0:13:38 – 0:13:45
It's less close to the real question. But it is highly interesting to deal with
|
|
|
0:13:45 – 0:13:56
to deal with this issue, because from a scientific point of view or from a scientific point of view
|
|
|
0:13:56 – 0:14:05
Perspective arises from this huge possibilities, chat histories, comments,
|
|
|
0:14:05 – 0:14:14
Search transcripts and just look for places that might be interesting.
|
|
|
0:14:14 – 0:14:21
For example, in hours of podcasts conducted at night,
|
|
|
0:14:21 – 0:14:27
could easily be used to find interesting and sensitive places again and again,
|
|
|
0:14:27 – 0:14:40
They are enlightening. Little stories, little stories, and that's easy for them.
|
|
|
0:14:40 – 0:14:47
Data is worth gold. Yeah, that's what I've been dealing with. Interestingly, these
|
|
|
0:14:47 – 0:14:56
models that perform these analyses. It's very cheap, because I think it's
|
|
|
0:14:56 – 0:15:08
Token costs 0.0000002 cents. So if you go and watch all the videos of the TV chat chest,
|
|
|
0:15:08 – 0:15:16
If you analyze the transcripts with it, you get a total price of less than 2
|
|
|
0:15:16 – 0:15:27
Dollar. You have to imagine that. And that's kind of like 1300 videos. That's so funny.
|
|
|
0:15:27 – 0:15:39
Yes, the request to ChatGBT and so on is much more expensive than that, if you use the
|
|
|
0:15:39 – 0:15:41
The programming interface.
|
|
|
0:15:41 – 0:15:48
And only for a short part of a long text.
|
|
|
0:15:48 – 0:15:52
Well, I'm such a tech nerd.
|
|
|
0:15:52 – 0:15:57
And this programming, that was so much fun for me,
|
|
|
0:15:57 – 0:16:02
Because they achieved good results so quickly.
|
|
|
0:16:02 – 0:16:06
And if you put the whole thing in a database,
|
|
|
0:16:06 – 0:16:14
which is accessible via the Internet and then also builds a website,
|
|
|
0:16:14 – 0:16:19
Anyone can actually go there and access this data.
|
|
|
0:16:19 – 0:16:23
I mean, this is a total brainfuck.
|
|
|
0:16:23 – 0:16:27
I mean, somebody could do that.
|
|
|
0:16:27 – 0:16:37
And then you could search contextually explicitly for interesting jobs.
|
|
|
0:16:37 – 0:16:41
And that could be done by anyone who has a knack for it.
|
|
|
0:16:41 – 0:16:45
It's kind of funny, isn't it?
|
|
|
0:16:45 – 0:16:48
Maybe someone should do it.
|
|
|
0:16:48 – 0:16:59
But back to surveys. Have you done it this way or are you doing it this way?
|
|
|
0:16:59 – 0:17:10
Pretty annoyed? Do you have time for a small survey? So I find road surveys
|
|
|
0:17:10 – 0:17:18
It's awful. I've never had time, I've never had time, and I'll never have time for that either.
|
|
|
0:17:18 – 0:17:26
Because I choose exactly where I do a survey. I find it really interesting,
|
|
|
0:17:26 – 0:17:34
In today's world, where you have unlimited access to data,
|
|
|
0:17:34 – 0:17:48
still the good old survey is such an important tool to provide quasi a snapshot for
|
|
|
0:17:48 – 0:17:58
Mood, interests, all sorts of reasons – I buy a car now when I
|
|
|
0:17:58 – 0:18:03
I would buy a car, what brand that would be, how much money I'm ready for a new car
|
|
|
0:18:03 – 0:18:10
To say that all these things are being raised and that they are still being raised in such a way.
|
|
|
0:18:10 – 0:18:18
1-to-1 situation. You could just use Big Tech and then just Big Data,
|
|
|
0:18:18 – 0:18:26
Because what Google and Co. know about us is so incredible. There is a link,
|
|
|
0:18:26 – 0:18:37
it's in the Google profile and if you click on it, I don't know exactly how the
|
|
|
0:18:37 – 0:18:44
It's still called, but I had just clicked on it some time ago and then it's right there,
|
|
|
0:18:44 – 0:18:51
What Google knows about you is incredible. If you look at it that way
|
|
|
0:18:51 – 0:18:58
Interests like this from the last search history, then all sorts of things appear.
|
|
|
0:18:58 – 0:19:06
And this is really very interesting, because from this Google concludes on various
|
|
|
0:19:06 – 0:19:12
Things. Let's just say that 99% is correct. So there you can see,
|
|
|
0:19:12 – 0:19:21
What these companies have accumulated with all their data collection for incredible knowledge.
|
|
|
0:19:21 – 0:19:26
And the interesting thing is, if I go now and I click on Google
|
|
|
0:19:26 – 0:19:33
"new computer case" because I'm looking for something, because I might have a small
|
|
|
0:19:33 – 0:19:47
Assemble servers for other tasks, such as creating vector data. Then I'll get up
|
|
|
0:19:47 – 0:19:58
Facebook, LinkedIn, and everywhere Computer Enclosure Ads for Server Computer Enclosures
|
|
|
0:19:58 – 0:20:06
for the rack and there you can see just how networked all this is and all the cookies that you
|
|
|
0:20:06 – 0:20:15
keeps planting in his browser, how they exchange all data and then the advertising networks
|
|
|
0:20:15 – 0:20:23
grabbing. It's all data that's collected somewhere. And this
|
|
|
0:20:23 – 0:20:31
It would be interesting if this were exploited instead of conducting such surveys. But
|
|
|
0:20:31 – 0:20:37
This, of course, is again the problem with data protection and data security. There we have
|
|
|
0:20:37 – 0:20:45
In Europe, it's all over the place anyway. If I consider before the General Data Protection Regulation
|
|
|
0:20:45 – 0:20:54
Being active, I was able to do much more with my website. Now it's all anonymized.
|
|
|
0:20:54 – 0:21:01
and encrypted and the data must be such that it can also be erased. And that's what makes
|
|
|
0:21:01 – 0:21:06
Keep everything a little bit more elaborate. But the basic idea behind it is that
|
|
|
0:21:06 – 0:21:13
Everyone should decide for themselves what happens to their data. But outside of Europe
|
|
|
0:21:13 – 0:21:25
The data is collected, not just on a massive scale, but in tons. You have to be
|
|
|
0:21:25 – 0:21:33
Imagine, I had seen a video of Semophois, this is a tech-youtuber,
|
|
|
0:21:33 – 0:21:41
He reports on interesting technological things with AI and what he programmed and
|
|
|
0:21:41 – 0:21:49
From time to time he also tells from sewing boxes. And he said he can't enter the U.S. anymore,
|
|
|
0:21:49 – 0:21:57
Because, you have to imagine, he has reported critically about Trump. And now that is so,
|
|
|
0:21:57 – 0:22:03
They don't just collect the data about the person who once criticized Trump.
|
|
|
0:22:03 – 0:22:13
But also for those who know him. And that's not enough, but those,
|
|
|
0:22:13 – 0:22:24
Those who know him. And that's unbelievably much. There is also this Kevin Bacon number.
|
|
|
0:22:24 – 0:22:32
That means that over six people know every other person in the world.
|
|
|
0:22:32 – 0:22:35
In the end, Kevin Bacon.
|
|
|
0:22:35 – 0:22:40
So I know someone who knows someone, who knows someone, and all six times
|
|
|
0:22:40 – 0:22:43
He knows Kevin Bacon.
|
|
|
0:22:43 – 0:22:45
That's wow.
|
|
|
0:22:45 – 0:22:48
Krass isn't it?
|
|
|
0:22:48 – 0:22:51
You have to think about that.
|
|
|
0:22:51 – 0:22:58
And that's why when they go there and almost the main person and then the one who knows
|
|
|
0:22:58 – 0:23:02
Someone who knows how many people they are.
|
|
|
0:23:02 – 0:23:05
That's unbelievable.
|
|
|
0:23:05 – 0:23:10
Yeah, that's it.
|
|
|
0:23:10 – 0:23:14
And that's where all the data is collected, everything you can find.
|
|
|
0:23:14 – 0:23:21
And since we use everything diligently American services, of course everything is in America.
|
|
|
0:23:21 – 0:23:27
stored. A scoundrel who thinks evil. Personally, I wouldn't go to the United States either.
|
|
|
0:23:27 – 0:23:36
They currently want to travel. It could be that I'm arrested at the check-in and I'm
|
|
|
0:23:36 – 0:23:45
You can have a great conversation with the officers there. I'd have less bucks, too.
|
|
|
0:23:45 – 0:23:52
on it. So there you can see what you can do so far away from surveys with such
|
|
|
0:23:52 – 0:23:59
data collection can be done. The question of whether there should be more autonomy in Europe
|
|
|
0:23:59 – 0:24:12
If you want to practice, that's really important. Digital sovereignty means that. And we should
|
|
|
0:24:12 – 0:24:21
And we also do better if we make ourselves as independent as possible. Unfortunately, you can
|
|
|
0:24:21 – 0:24:29
Some things just don't replace it. The offerings in artificial intelligence that are available in
|
|
|
0:24:29 – 0:24:39
Europe is not as good and at the same level as the US.
|
|
|
0:24:39 – 0:24:41
as well as Chinese products.
|
|
|
0:24:41 – 0:24:45
But I'm hoping this will come.
|
|
|
0:24:45 – 0:24:50
And for all other things, there are now enough alternatives.
|
|
|
0:24:50 – 0:24:54
So instead of Cloudflare you can
|
|
|
0:24:54 – 0:24:59
Bunny CND,
|
|
|
0:24:59 – 0:25:05
Instead of Gmail or Gmail, you can use Proton Mail.
|
|
|
0:25:06 – 0:25:14
And there are so many other European providers who also do nothing in US data centers.
|
|
|
0:25:14 – 0:25:17
You have to find out.
|
|
|
0:25:17 – 0:25:23
The Morpheus has an interesting project and they are building a database with European
|
|
|
0:25:23 – 0:25:26
Alternatives to U.S. products.
|
|
|
0:25:26 – 0:25:34
And of course there are also many free software or services that, in places,
|
|
|
0:25:34 – 0:25:41
They are expensive, even cheaper, like the American ones. If you look at it,
|
|
|
0:25:41 – 0:25:47
If you are interested in it. So for me it's a very important thing,
|
|
|
0:25:47 – 0:25:59
Because my data is important to me. All right, did I leave today? A little bit. me
|
|
|
0:25:59 – 0:26:03
I've talked a bit too much about vectors and what you can do with them.
|
|
|
0:26:03 – 0:26:10
Okay, I'll say thank you and see you next time.
|
|
|
0:26:10 – 0:26:11
Goodbye, Sasha.
|
|
|
0:26:11 – 0:26:13
[Music]
|
|
|
0:26:13 – 0:26:16
[Music]
|
|
|
0:26:16 – 0:26:22
[Music]
|
|
|
0:26:22 – 0:26:28
[Music]
|
|
|
0:26:28 – 0:26:34
[Music]
|
|
|
0:26:34 – 0:26:40
[Music]
|
|
|
0:26:40 – 0:26:43
[Music]
|
|
|
0:26:43 – 0:26:49
[Music]
|
|
|
0:26:49 – 0:26:55
[Music]
|
|
|
0:26:55 – 0:27:01
[Music]
|
|
|
0:27:01 – 0:27:08
[Music]
|
|
|
0:27:08 – 0:27:11
[Music]
|
|
|
0:27:11 – 0:27:17
[Music]
|
|
|
0:27:17 – 0:27:23
[Music]
|
|
|
0:27:23 – 0:27:29
[Music]
|
|
|
0:27:29 – 0:27:35
[Music]
|
|
|
0:27:35 – 0:27:38
[Music]
|
|
|
0:27:38 – 0:27:44
[Music]
|
|
|
0:27:44 – 0:27:50
[Music]
|
|
|
0:27:50 – 0:27:56
[Music]
|
|
|
0:27:56 – 0:28:02
[Music]
|
|
|
0:28:02 – 0:28:05
[Music]
|
|
|
0:28:05 – 0:28:11
[Music]
|
|
|
0:28:11 – 0:28:17
[Music]
|
|
|
0:28:17 – 0:28:23
[Music]
|
|
|
0:28:23 – 0:28:29
[Music]
|
|
|
0:28:29 – 0:28:32
[Music]
|
|
|
0:28:32 – 0:28:38
[Music]
|
|
|
0:28:38 – 0:28:44
[Music]
|
|
|
0:28:44 – 0:28:50
[Music]
|
|
|
0:28:50 – 0:28:56
[Music]
|
|
|
0:28:56 – 0:28:59
[Music]
|
|
|
0:28:59 – 0:29:06
[Music]
|
|
|
0:29:06 – 0:29:12
[Music]
|
|
|
0:29:12 – 0:29:19
[Music]
|
|
|
0:29:19 – 0:29:22
[Music]
|
|
|
0:29:22 – 0:29:28
[Music]
|
|
|
0:29:28 – 0:29:34
[Music]
|
|
|
0:29:34 – 0:29:40
[Music]
|
|
|
0:29:40 – 0:29:46
[Music]
|
|
|
0:29:46 – 0:29:49
[Music]
|
|
|
0:29:49 – 0:29:55
[Music]
|
|
|
0:29:55 – 0:30:01
[Music]
|
|
|
0:30:01 – 0:30:07
[Music]
|
|
|
0:30:07 – 0:30:13
[Music]
|
|
|
0:30:13 – 0:30:31
[Music]
|
One goes unsuspectingly to the mailbox – and suddenly there is a letter from a federal government institution. In it: a small envelope, a QR code for a survey and actually five euros as a thank you. Quite a battered five, but at least.
From this small incident, a typical digression develops: How do surveys actually work? Why are questions often asked twice? And how reliable are such data collections?
From there, the thought quickly leads to a much larger topic: data. While traditional surveys survey people directly, large platforms in the background have long since collected huge amounts of information about interests, behavior and decisions. What used to be done with questionnaires today often happens automatically via search queries, cookies and user profiles.
And then it gets nerdy: What happens when you search texts not only for keywords, but for meaning? Context-based search with vectors makes it possible to find content according to its meaning – not just individual words. An exciting look at how modern AI systems can analyze texts, search transcripts or recognize correlations in huge amounts of data.
An episode about surveys, data collection, artificial intelligence and how much we actually reveal about ourselves – sometimes voluntarily, sometimes unnoticed.
And, of course, the most important insight of the day:
Apparently, my opinion is worth exactly five euros.
Topic overview:
The letter in the mailbox
An unexpected letter from a state institution – and a €5 note as a thank you for a survey.
Why Surveys Ask Questions Doubly
How questionnaires are structured and why some questions seem to be asked identically.
My Experience with Scientific Surveys
Podcast surveys, community studies, and the scientific value of small samples.
When surveys make sense – and when not
Reputable institutes vs. data collectors and why you should be careful.
Big data instead of questionnaires
What big platforms already know about our interests.
Tracking and advertising networks
Why suddenly, after a search, suitable advertising appears everywhere.
Nerd excursion: Context-based search with vectors
How modern systems search texts not only by words, but by meaning.
AI, databases and huge transcripts
Why this technique is particularly exciting for podcasts, comments and large text collections.
Digital Sovereignty in Europe
Why alternatives to major US services are becoming increasingly important.
Conclusion of the digression
A small survey, a wrinkled five – and surprisingly many thoughts about data, AI and our digital world.
Links to the issue:
Advertising:
Contributors to this episode
Sascha
Playing time: 0:30:32
Date of admission: 05.03.26
This episode is from season 4, there are a total of 4 seasons.
One According to Funk Publication Cartel Production - Where stories are transmitted unmistakably.
Podcast license:
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License
Note: This post contains affiliate links. If you shop through these links, I'll get a small commission at no extra cost to you.
Related posts
Discover more from LautFunk Podcast & Blog
Subscribe to get the latest posts sent to your email.
Author: Sascha Markmann
Sascha Markmann is a creative mind with a moving biography – and a person who has rediscovered life after a stroke. After working as a paramedic and elderly caregiver, Sascha found his way into creative expression. As a blogger, podcaster, musician and visual storyteller, he brings together things that just don't fit at first glance: He combines his keen enthusiasm for technology – from self-hosted servers to electronic music production – with emotional depth and an oblique sense of humor. His contributions are somewhere between borderline, acid basslines and assistance – honestly, directly and with a wink. Guiding principle: “Audio-visual dullness with no usefulness – but perhaps that's why it's so valuable.” View all posts by Sascha Markmann
English