EU chief calls for restriction of Internet giants' control over freedom of speech

Hanno Labuschagne

Journalist
Staff member
Joined
Sep 2, 2019
Messages
5,339
EU chief calls for restriction of Internet giants' control over freedom of speech

Restrictions should be placed on internet companies so they aren’t in a position to make unilateral decisions that will affect freedom of speech and democratic values, European Commission President Ursula von der Leyen said.

“We need to contain this immense power of the big digital companies,” von der Leyen said in a virtual address at the World Economic Forum.

“We want the platforms to be transparent about how their algorithms work because we cannot accept that decisions that have a far-reaching impact on our democracy are taken by computer programs alone.”

[Bloomberg]
 
She said it wasn’t for internet companies alone to decide how information was disseminated.

Of course it is. The democratic narrative is the only acceptable narrative. Everything else is therefore racist, fascist and undemocratic.
 
Those algorithms are by human design, it isn't computer controlled.

Sounds like a excuse FNB once gave to me, it is not our error it is a system error.
It depends on what we're talking about.
If it's ML-based algorithms then no, the programmers literally have no clue how they work internally.
Google Search (RankBrain) for example is ML-based. They can train the AI, but they can't tell you definitively how it arrives at its results.
 
It depends on what we're talking about.
If it's ML-based algorithms then no, the programmers literally have no clue how they work internally.
Google Search (RankBrain) for example is ML-based. They can train the AI, but they can't tell you definitively how it arrives at its results.

They sure do sampling which is human reviewed for algorithm correction.
 
It depends on what we're talking about.
If it's ML-based algorithms then no, the programmers literally have no clue how they work internally.
Google Search (RankBrain) for example is ML-based. They can train the AI, but they can't tell you definitively how it arrives at its results.
White box machine learning algorithms do exist.
 
I totally agree with this. These companies should be restricted in their operations and controlled by governments globally where they operate. There should be standard rules all governments need to adhere to when ensuring they operate within the boundaries set. Time to break these companies up, take away any power or authority they may have and leave them focussing on their core businesses. If they do not like it, close them down permanently.
 
White box machine learning algorithms do exist.
But not for Google search, which is the example I used?
So how does Google align with this new possible EU imperative?

Performant machine-learning at scale depends on deep-learning. Deep-learning by definition is not white-box.
You think Facebook, Twitter, and others are not leveraging DL to deliver their products?
 
It depends on what we're talking about.
If it's ML-based algorithms then no, the programmers literally have no clue how they work internally.
Google Search (RankBrain) for example is ML-based. They can train the AI, but they can't tell you definitively how it arrives at its results.
Not completely true. Programmers still decide how it arrives at the algorithms. It's not like you throw a bunch of random vegetables in a pot and see what comes out. I've said it before, there is no real AI, it's all just illusion to appear like a computer is making decisions.
 
But not for Google search, which is the example I used?
So how does Google align with this new possible EU imperative?

Performant machine-learning at scale depends on deep-learning. Deep-learning by definition is not white-box.
You think Facebook, Twitter, and others are not leveraging DL to deliver their products?
Whilst deep learning may not be white-box, it's greatly influenced by the human bias introduced through the training model... feed the model enough 'burgers good, vegetables bad' type inputs and the machine will eventually prefer burgers over vegetables.

D
 
Whilst deep learning may not be white-box, it's greatly influenced by the human bias introduced through the training model... feed the model enough 'burgers good, vegetables bad' type inputs and the machine will eventually prefer burgers over vegetables.

D
Finally. Now, would the EU accept not knowing exactly how a system is doing that?
How can a company be transparent about a system that's by definition opaque.
My point is that, will companies have to change their systems to accommodate the EU?
 
Finally. Now, would the EU accept not knowing exactly how a system is doing that?
How can a company be transparent about a system that's by definition opaque.
My point is that, will companies have to change their systems to accommodate the EU?
It isn't about changing systems. Facebook has in recent months made a lot of human decisions over what information is allowed on their platforms. They have too much power and influence in the world.
 
This is going to cause cognitive dissonance for progressives.
That doesnt really make sense. Arent you complaining all the time about "socialism"... and now when the government wants to regulate the corporations more, all of a sudden your free market fundamentalism / "small government" vibe goes out the door?

I dont think its the progressives that will have a problem dealing with this.

I consider myself a progressive and I have long believe companies need to be regulated smartly to promote a free and fair market.
 
Last edited:
It depends on what we're talking about.
If it's ML-based algorithms then no, the programmers literally have no clue how they work internally.
Google Search (RankBrain) for example is ML-based. They can train the AI, but they can't tell you definitively how it arrives at its results.

It wasn't an algorithm that recommended the New York Post story about Hunter Biden's laptop be blocked and the NYP's account suspended. It was a human.
 
  • Like
Reactions: Swa
But not for Google search, which is the example I used?
So how does Google align with this new possible EU imperative?

Performant machine-learning at scale depends on deep-learning. Deep-learning by definition is not white-box.
You think Facebook, Twitter, and others are not leveraging DL to deliver their products?
Doesn't affect Google as much as it affects Twitter and Facebook.

Twitter and Facebook employ content moderators who decide what content is acceptable or not - this needs to be done in a more transparent fashion.

You have cases like the Hunter Biden incident mentioned above, that had a real impact on an election, that both Twitter and Facebook suppressed.

The important point is that neither Facebook nor Twitter are equipped or qualified to judge whether or not such a story is true. In fact, since the FBI is investigating, it appears that it is at least partially true.
 
Back
Top