Stellar Community Fund

By ICanProveIt -Proof-of-Learning

ICanProveIt -Proof-of-Learning

SCF #26

$50.0K

Legacy v5.0 Activation Award

Awarded

An AI exam generation platform for the creation of universally respected, academic-standard proof-of-learning certificates.

Products & Services

ICanProveIt is a digital certification platform that uses advanced technologies to issue verifiable, blockchain-based educational certificates. The platform combines AI, blockchain (Stellar), and decentralized identity (DID) to create proof-of-learning certificates that adhere to high pedagogical standards accepted universally by academic institutions. ICanProveIt is designed by university professors and academics to not be adversarial to the academic world. While ICanProveIt may challenge traditional educational roles it also promises to liberate teaching professionals from administrative burdens, enabling them to focus more on teaching and research. ICanProveIt was designed to be teacher/professor friendly.

This platform addresses multiple challenges:

Non-Demonstrable Knowledge: A significant issue is that 98% of our knowledge and capabilities are non-demonstrable and unshareable. ICanProveIt creates a digital, verifiable format that allows individuals to demonstrate and share their learning effectively.

Chaotic Proof-of-Learning Outside Academia: Outside of the academic world, proof-of-learning is often chaotic, pedagogically repulsive, and largely ignored. The platform standardizes the validation process with rigorous academic oversight to restore credibility.

Limited Formal Education Scope: Formal education covers less than 1% of subjects that people learn throughout their lives. ICanProveIt expands the scope by certifying a broader range of knowledge and skills, regardless of how they were acquired.

Lack of Verifiability: Traditional online learning often lacks a reliable method to verify educational achievements. ICanProveIt solves this by using blockchain technology for immutable record-keeping.

Education Accessibility: High costs and geographic barriers often limit educational opportunities. The platform provides affordable, universally accessible educational certification.

Standardization: There is a disparate recognition of educational certificates across different geographies and institutions. By adhering to globally recognized standards, ICanProveIt ensures its certificates have wide acceptance.

Mass Retraining Needs Induced by Job Markets Being Disrupted by AI: The platform also aims to assist those whose jobs have been disrupted by AI, offering a means to retrain and certify new skills in a rapidly changing job market.

The platform primarily serves self-learners, educational institutions, and employers. Self-learners benefit from obtaining verifiable credentials for informal learning, institutions can reduce administrative costs and maintain integrity in credential issuance, and employers receive reliable proof of candidates' learning.

ICanProveIt embraces continuous learners and recognizes that self-learning has real value to employers, and the employment market. The internet provides us with nearly unlimited opportunities follow our interests and passions. Learners can pick from blogs and YouTube videos or take fee courses from the world’s top universities given by the professors famous for their contributions or their teaching abilities. But there is a catch! Without a reputable certificate as proof-of-learning, the learner is unlikely to receive immediate benefits from his learning. No employer is going to hire someone on the basis of the learner claiming to have read a book or taken a free course with nothing to prove it. ICanProveIt works in 3 steps: The Learner follows their normal path to acquiring knowledge by reading books and blogs, watching videos and taking free online courses from universities. The learner uploads or provides us with links to their learning material. ICanProveIt platform creates an exam by matching the provided learning material against material in a repository curated by academics ensured to be up to date. After successfully completing the exam the learner is issued a certificate that can be put on a resume or shared on social media.

AI Integration: AI is used to generate personalized exams based on the content learned by users, ensuring that the assessment is tailored to the individual’s study material. Blockchain (Soroban): Stellar’s blockchain technology offers a decentralized, fast, and low-cost ledger to store certificate data, making the issuance and verification of certificates secure and efficient. Soroban Smart Contracts: Utilizing Soroban, Stellar's smart contract platform, allows for complex operations like conditional certificate issuance and automated verification processes. This enhances the scalability and functionality of the certification process.

In contrast to the typical, rigid model, where learners are each given the same set of study material, followed by the same exam, ICanProveIt tests learners on the specific material they have consumed, including material from YouTube videos, blog posts, traditional or audiobooks etc. All exams are created on the fly from a curated repository of content matched with the learners' uploaded content. In the case of courses that mimic those found at traditional universities that have a generally recognized curriculum, the learner’s uploaded content will not be used for exam creation but instead will be used to inform the learner as to whether they have sufficiently studied enough (assuming that they studied all of the material uploaded) to be successful in the examination. Learners will still be able to take tests on such standardized classes if their material does not completely satisfy this requirement, but required areas are not covered by the uploaded materials and will draw their questions from a set of curated materials that reflect the standards required for passage of the material. Generating exams that are relevant to employers means following pedagogical standards that allow for comparability across subjects. Most online learning certificates are deemed worthless simply because there is little to no information about the test or material studied. Hiring managers are unlikely to dive deeper into the specific areas studied, the source of the information or the reputation of the certificate issuer. ICanProveIt is designed so that all exams generated are relevant to the learning goals and are formulated in a coherent and logical manner, one that clearly demonstrates the acquired knowledge of the student.

Coverage of Source Material - A metric ranging from 0 - 1 that indicates how much of the input text (material studied) is reflected in the quiz. It is calculated by mapping each question to the relevant sentences in the text and comparing the length of the mapped text to the total passage length. This method is based on the pyramid method used in summary annotations. Coverage of Curated Material - In cases where there is a typical course content, such as is the case for most principles and intermediate courses in various subjects at the world’s universities, the source material coverage metric will be replaced with the curated material coverage metric that covers the entirety of what is normally expected from a student in such a course and that is similarly given a metric ranging from 0 -1 as evidenced in the Coverage of Source Material metric. This prevents students from selectively choosing material that is narrower than what would typically be found in a standardized course. Employers or associations that wish to ensure coverage of specific topics can have standardized courses offered and learners will be notified as to what percentage of curated material is covered by their source material so that they can determine whether they wish to proceed with the standardized examination or whether they wish to read more of the recommended curated material instead.

Structure - A 1-3 metric assessing if a set of questions flows logically together, i.e. from easy to difficult, or that natural chronological order is used. For example, if the student was being examined on the formation of the universe, the questioning would start with the Big Bang and work towards the creation of our Solar System. This is similar to a measure used in conversational QG, where questions are logically linked for natural conversation (Mulla and Gharpure 2023). Redundancy - A 1 to 3 metric assessing repetition within an exam, such as identical questions being asked that do not require a differing perspective or adaptation of the student's thought processes. This has been utilized in conversational QG to prevent repetition and ensure natural conversation (Mulla and Gharpure 2023).

Ensuring a heightened level of quality regarding the entire test is crucial. However, the individually tailored questions must also meet exceptionally high standards. Questions must be rated on three metrics used to measure the fundamental aspects of a question's quality: Relevancy - A binary metric which measures whether the question is semantically relevant to the input context. Fluency - A binary metric used to assess the grammatical correctness and clarity of language in a set of questions. This metric is employed by Mazidi and Nielsen (2014) and Elkins et al. (2023), and is also used in different scales in Mulla and Gharpure (2023). Answerability - A binary metric which measures whether the question can be answered from the input context. It is not necessary to be able to find a passage from the input that is an answer to the question; it is enough if a student could reasonably answer the question from the context. For example, applying logic explained in the passage to a new situation makes the question ‘answerable’. As above, previous work by Steuer et al. (2021) and Elkins et al. (2023) uses a similar binary metric, and Mulla and Gharpure (2023) suggest similar metrics on different scales.