I'm real.

I'm Real - Ince Choi Lane: Concept + Identity

There are five levels to verifying information when it comes to the internet. From verifying hardware to verifying the individual. (The system I use goes as following: 1a. Affirming Device Capacity, 1b. Affirming Device Specification, 2. Affirming Humanity, 3. Affirming Skillset, 4. Affirming Identity, 5. Affirming Purpose). The current industry is focused on the extremes, which is why I want to help bridge the gap. Using minimal context to prove the most basic thing, that I am human (i.e. 1a to 3), alongside coming up with a few ideas about the higher tiers (i.e. 4 to 5, things like affirming membership and individual uniqueness).

With the development of AI and LLMs, we constantly need to prove that we are real people and that we are the right person for the job. The industry is currently split in half, they either have overly easy captchas or overly complicated captchas. The problem is that they are testing human intelligence. That is easy to mimic, in any level. What we need are physical things, our behaviour and practices.

If I were to state an explicit group affected, I suppose that its everyone. It serves as an alternative to traditional ways of human verification that are lower stakes. Not everything needs ten levels of security.

The idea is that the user would store these niche pieces of context that are as minimal and unobtrusive as possible. People are very protective of their data, so let's just store data that they would not feel the need to protect. For example, we can show the person a digital fidget spinner, get 10 volunteers to interact with it so we have a training model, and compare it to people who interact with it. An AI can draw curved or jittery lines, but it won't unprompted. This uses human behaviour to separate humans and AIs, being less intrusive than something like a captcha. Which can be infuriating or intrusive. These new ideas cover tiers two to three.

Being somewhat low stakes, the main concern is that it can't guarantee a 100% chance that it is a real person. This is why the lower tiers should be used to bypass low to medium tier captchas for showing skills or humanity, but not high tier things like personal identity or job hunting.

It is this issue that makes the higher and lower tiers more difficult to implement. This is why tier one is split into two. As though proving the exact model of a device requires the device's build information, proving a device's capabilities can be implemented in less intrusive ways. Which I have suggested to be running increasingly complex pieces of code to see how much the computer can handle, which the exact level of complexity would be stored and be used to see if a computer can run a certain program.

The most difficult tiers to implement within this affirmation tier system are tiers four to five. As you may often require security biometrics and physical identification to affirm one's personality or reason to do a certain thing. Which is why we have thought about ways to reduce the biometric data taken from the user. Such as instead of scanning one's face, we capture the dimensions between the features. The facial gaps between every human are unique, and the information we capture is not enough to reform one's facial features. Therefore affirming your individuality without requiring intrusive data. For affirming one's purpose (tier five), you would probably need something that is somewhat exclusive. For example, if we used the inspiration example with the approving one's membership, maybe we could have a qr code on a certificate that when scanned affirms that the certificate owner is certified for a certain job when paired with the prior biometric gap system to affirm that the person accessing the data is or is approved by the certificate owner.

The user controls access. They can delete it like any other form of data, or revoke it, or redo it, or change it. It should be similar to other data, just with behaviour or characteristics instead of information. With more information being given as the project requires a higher tier of affirmation.

Ultimately, it is simple. Utilise my five tier framework to decide what level of security whatever the project requires, and allow the user to input unobtrusive data to affirm the characteristics needed. Here is the framework performed on the four examples you have given us in the powerpoint.

Problem: human checks often demand legal identity. Passport solution: prove humanness without showing legal ID. My framework: tier two, use human interaction with the digital fidget spinner to distinct between the real and artificial.

Problem: eligibility checks ask for extra records. Passport solution: prove only the required fact. My framework: tier three, use the ABCDEFG keyboard to make typing more difficult, slowing down the user so that the user types slower than bots who are trained on a human's normal typing speed with QWERTY, but faster than someone directly copying from an AI on another device. After which we use an LLM to test that the logic makes sense, affirming someone's skill set.

Problem: communities need one person per account. Passport solution: verify uniqueness without public names. My framework: tier four, map a general facial structure with points on one person's face. Because everyone's face is different, there cannot be a repeat of one's facial structure. This is also less obtrusive than a face scan as you cannot reform a person's facial features from dimensions.

Problem: affiliation can expose private context. Passport solution: prove membership while hiding unrelated details. My framework: tier five, use a particular QR code on the certificate which is printed on which is unlocked by a person's general facial structure. When approved, they can forward this to any job recruiters with a time limit so that it affirms that they are the individual, have the qualifications, and are forwarding their information to someone else.

The video explains most of it in more detail. -> https://youtu.be/dNg8P_8SXK8

Built With

• canva
• google-drive
• me
• vn-video-editor