WEEKLY FINDINGS IN DATA SCIENCE AND AI

Heaven Banning

You've probably heard about shadow banning. With shadow banning, users are not kicked out of the community. They can continue to use their accounts and post, but no one will see their content. They are not told about this change, thus, to them, it might look like they are not banned at all.

A more recently talked about option is called heaven banning. Just like shadow banning, the user is not told of the ban and the posts are not shown to other users. The difference is that the user will afterward only interact with bots/AI that will echo back positively. George Hotz, who is currently working for Twitter, brought up this method and argued it could reduce toxicity on the internet:

But I wholeheartedly disagree, in fact, I think it's extremely immoral and evil. First of all, people will never understand their mistakes if they are positively met by the bots. When they are shadow banned there is at least an indication that their content is bad, since they won't get likes or replies.

Secondly, they are essentially put into a mental prison. They could waste a lot of time talking to these bots, especially if they are powered by AI. I don't think wasting people's time is like that is okay. Life is time, and taking it is wrong. You could argue that if they were banned they may have done something wrong, but still, it's likely not to the level of actually going to prison.

ChatGPT

The new language model ChatGPT has gained a lot of attention due to its impressive capabilities, and it seems like every day that passes something new has been discovered about it. This process was accelerated due to the creators, OpenAI, making it free for everyone to use. I've written two articles about it.

An introductory article:

And a method for improving the responses:

One of the recent discoveries that is causing a lot of buzz recently is instructing the AI to emulate a terminal/virtual machine:

In reality, it's not executing commands but using what it has learned to simulate or "hallucinate" realistic responses given the context.

Updates From Neuralink

Neuralink is a company led by Elon Musk that is developing a brain-computer interface using a small implanted device called Link to capture brain waves. These signals can then be processed using deep learning and used for various tasks.

This week an updated presentation of their progress was made. Elon Musk talked about humans already being "cyborgs" since our phones are part of our everyday life. The weakness, he argues, is that sending and receiving information is very slow using phones. In contrast, if we had access to a brain-computer connection, the time could be decreased considerably.

The technology has been tested on monkeys and successfully enabled them to do different tasks such as playing games and typing using only their mind:

Now, Neuralink has submitted information to the FDA to get approval for testing on humans. He predicts that in 6 months the first human could be tested. One application is restoring vision and supposedly it could work even if the person is completely blind!

I have mixed feelings about Neuralink. In terms of positive aspects, the restoration of vision would be incredible, and the potential increase in productivity would be a game changer. Also, if we have access to AI directly through our brain, we are essentially AI + Human, the best of both worlds. Consequently, all updates to AI would benefit us, and we won't be surpassed by it.

That being said, it seems quite risky to have a device constantly active in our heads. It would be crucial that privacy and security are top-notch. Additionally, I wonder if we can ever step away from technology if it's implanted inside our brains. One could argue that the fact that phones are as integrated into our everyday life as they are isn't necessarily a good thing.

For now, it's not clear how far Neuralink can go. We just have to wait and see.

If you're interested in reading more articles about data science or AI, check out my reading lists below:

If you'd like to get a Medium membership you can use my referral link if you wish. Have a nice day.