Hello all, see below for my elevator and venture pitch
Elevator:
Venture:
A3: AiMarker – An adaptive AI that marks your assessments
12 responses to “A3: AiMarker – An adaptive AI that marks your assessments”
Leave a Reply
You must be logged in to post a comment.
As a teacher, I would love to see something like this help me mark essays. I can also see where ‘teaching the system how you mark’ could also help. But do you really think that you could teach an AI that much in 5 years and be functional… (and for which subjects? Essays? Math?)?
I think it begins with standardized tests like AP and A Level exams, before branching off to more classroom specific implementations. The key characteristic of AI is that it self-optimizes. It teaches itself using available information. AI today can already interpret meaning from language and perform tasks as a result. Collegeboard has decades of past papers and data it can feed into an AI and the AI will be able to learn the marking habits of AP markers in each specific course. It’ll be able to mark even more objectively than different human markers. 5 years was indeed an imagined timeline and probably rushed. But it’s going to happen, I’m sure of it.
FEEDBACK: the concept is indeed compelling, however, its success is still unproven as one thing AI cannot do effectively is grade and provide feedback for essay and open questions. This is uniquely a human task. I would have loved to learn about the business model and the ROI, the ask, and the marketing strategies.
I used to believe that writing was uniquely human, and that might be true today for original writing. However, for reproductions of existing information, for writing that needs to be mapped to a standardized rubric, AI will be more than ready to do so, and self-assess, in the next few years. Today, many blog articles and YouTube videos are already AI-generated. I encourage you to go test the writing capabilities of OpenAI. It’s only 2022, and it’s getting smarter so quickly it’s actually kind of scary.
FEEDBACK: Hi Leon, your venture idea is an interesting one. As a teacher, assessment is one of the most grueling parts of the job, so having AI assist in that department is a wonderful and feasible idea. There are already AI platforms that support individualized learning and support teachers in lesson planning, so the direction of your venture is on point. Having been one of those manual graders for standardized testing, I can see how outdated, laborious and costly it is versus using a tool like AImarker. Thus, I like how you mentioned a partnership between big board exams. I appreciate your last quote in your elevator pitch “to make teaching the main part of the teacher’s job again” especially with your statistics to back up the number of hours spent on administrative tasks that can be completed by AI. I would have liked to have a better understanding of the market and the competitors, even indirect competitors that share similar features as well as the projected market for this software. There are still a few unknowns about the technology used to power this AI software, thus I would wait to invest on this venture.
For sure, AI marking is still a new frontier waiting to be explored. I think partnering with big board exams and using machine learning to learn how to mark more standardized written responses will be a good springboard for AI marking in classroom specific implementations.
FEEDBACK: Very interesting venture idea Leon! I am curious about the potential of AI to do this successfully – I wonder what the options to learn to grade and provide personalized feedback suggestions based on the teacher or institution’s past suggestions would look like. I think the grading (if multiple choice), plagiarism checker, and analytics for potential targeted skills to inform future instruction all seem feasible. I LOVE the idea, I’m just not sure if AI is at the point where it can be done effectively. Education is moving toward creative generation and critical analysis, which can look so many different ways translating into many “right” answers. These skills would be difficult to accurately assess using AI, which might miss out on the nuances.
MC grading, plagiarism checking and analytics for targeted skills already exist so I wouldn’t think of these functions as revolutionary. They’re more accessory to AI marking. Today, AI is at the point where it can write blog articles. From AI art to AI videos, we already know that an AI has the capacity to creative in ways that humans can learn from. I think as AIs get more sophisticated and increase their ability to understand and decode interactions, interacting with an AI online will eventually be indistinguishable from interacting with a human online, except an AI has access to all publicly available information in a moment’s notice. It will be like having a conversation with Google if Google was a person.
A little dystopian and hard to understand, but it’s where we’re headed. Google is already testing AI receptionists that are indistinguishable from human receptionists via phone. AI today can already answer existential philosophy questions. We’ve made much progress in a few short years, who knows where we’re headed.
FEEDBACK: Hi Leon, AiMarker faces the challenge of the teacher’s professional purpose and the time of the work routine. The ratio of teaching time to time spent on evaluation and analysis is impressive. Indeed, we need tools that streamline routine processes to focus on teaching. The same conditions are repeated in Health professionals.
Nice job!
FEEDBACK: Hello Leon, AiMarker is an interesting idea, but I do wonder how much time we would consume for step #4 “Review and make suggestions to AIMarkers feedback and scores as needed before final approval”. When I correct I don’t have to review my feedback because I anallyzed the exam, corrected it and then write the feedback when there is a mistake. I would use your product, but it depends for what subject I am assessing. For instance in English, if I am evaluating an essay, I would want to read it myself and then provide the feedback, but for many other assessments this AiMarker would be great. I enjoyed watching your elevator pitch and venture pitch. Thank you!
Review/Feedback: Hi Leon, your elevator pitch was clear and well presented, but I would encourage you to make it flashier/eye-catching. You have 1 minute to grab attention, I think your elevator pitch could have been a little more passionate or enthusiastic with how amazing AIMarkers is. I would challenge to to add more graphics and eye catching stats to your elevator pitch. Your venture pitch is incredibly thorough and well-researched. You focus on the 3 markets AI Markers reach, but I think you could focus more on competitive advantage and comparisons. Overall, it seems like a great venture, thank you for sharing!
FEEDBACK: Hi Leon, thanks for your venture pitch. You did a great job at explaining the problem that educators face when it comes to marking assessments that are more abstract than typical multiple choice. Your venture pitch was very insightful, and it was evident that you’ve put a lot of thought and time working through the various features educators would require. One thing I’m often skeptical about when it comes to AI is the training data that it’s based off of not being diverse enough. We often see stories regarding how specific minority groups were left out of the training data and adversely affected when the AI programs were used in production (eg. AI being tested to determine jail time for crimes skewing toward longer sentences for individuals who are black), how will AiMarker avoid these kinds of biases when grading students work, especially when students’ lived experiences can greatly impact their interpretations of essay style questions. Thanks again for your excellent venture pitch!