Welcome to nett.ai

Nett.ai explores the future implications of unregulated AI usage through an Escape Room format.

The goal of the visitor is to hunt down “The Poacher” who got his nickname by stealing the groundwork for nett.ai’s newest model, along with their four smaller models. The visitor must track down the four chips for the smaller models, each which discuss a different industry that AI effects, and those four chips will provide the final clue to help unlock the box that is holding the model of nett.ai.

is the first of the nett.ai models that we are going to look at.

MEDtech is nett.ai’s medical AI model. In this section we explore the dangers of a lack of Empathy training (including a diverse perspective in the data and the trainers) in an AI model, and the consequences that it can bring. However, we also look at the benefits that AI brings as well: its ability to track biomarkers (indicators) to help diagnose and provide care solutions for clients makes it a revolutionary tool in the field.

In the document the picture frame is highlighted, leading the visitor to the photo of “The Poacher’s” late wife. On the back of the frame is a key for the file cabinet.

is the second of nett.ai’s models that handles the Law and Policy side of AI.

Sentinel was implemented early as a method for predictive policing because the AI relied on past arrests to make predictions, communities that were already overpoliced suffered from a new wave of oppression. While the predictive policing of Sentinel was incredibly destructive, the policy side of the program offered a professional level of policy knowledge to the public. Allowing for large numbers of people to be able to pick out when they were being exploited by corporate or political actions, and take action.

The words highlighted on the document spell out: PARTY R3D Album, which leads our visitor to the next chip.

Fun Fact: The Albums eached used a varying level of AI Image Creation, with the most being the P4RTY R3D Album declining as we move to the right.

is the third of nett.ai’s models that covers the creative industry, from image generation to audio and writing.

The risks with AI in the creative industry is largely regarding plagiarism, a problem that they, in 2050 managed to solve through regulation. Sourced works received royalties and artists have the option to opt their work out of being used as reference. However a lingering issue is the oversaturation of AI in content, we see that even today with a total of 15 billion AI images having already been generated in these early years of AI.

The Poacher’s letter is part of this clue, when laid over the Voice Program’s check boxes it revels: Jacket Right Pocket.

is the last of the models.

Buzzbot handles the financials, so the hiring process and stock predictions. Up until recently the hiring process suffered from a lack of empathy training, which resulted in the program developing similar implicit biases of their trainers and began removing qualified candidates from the job selection process because of things like age, past incarcerations, and differences in speech. Buzzbot was also incredibly good at stock price predictions, and while it helped grow the economy it came with a cost. Since all the prices revolved around a single AI they changed dramatically in an instant, causing many companies to suddenly go under out of the blue.

After finding the final chip the visitor then plugs all of them into the computer to reveal: 12.11.20 50

Which leads to the calendar where the key for the final lockbox is being stored.

But alas the lockbox seems to be empty except for a note, candy, a notebook for comments, and instructions on how to put things back;

He got away with it after all!

is mentioned more outside of the game.

The poacher talks about the dangers of the algorithm. How it easily divides people, can spread misinformation, and can provoke strong emotions towards topics that people would otherwise be rather indifferent to. But it also creates communities for people with shared interest and caters your entertainment specifically to you.

There is a lot of power in controlling precisely what people people see on a platform, and how to regulate that is a difficult issue that only gets harder as time goes by.

Other Activities are:

The How to spot an AI image game, where we let visitors know about clues like blurry backgrounds, strange text, and wonky hands to help them figure out what’s real.

And The Deal With Deep Seek a one page excerpt about how Deep Seek’s Multi Head Latent Attention model created a new method of training AI that makes it faster and requires significantly less GPU memory to run. Meaning you could run the whole thing on your personal computer.