- Brain Scriblr
- Posts
- Using AI for Tutoring
Using AI for Tutoring
Chester Beard
January 21, 2025
News
Before leaving office then President Joe Biden signed an excutive order directing the DoD and DoE to select (and then lease) sites where the private sector can build and operate “gigawatt scale” AI data centers, as well as clean power facilities, all intended to minimize negative impacts on local communities and the environment.
An initial hearing on the suits New York Times vs OpenAI was held last week. No decision yet if the case can move forward or not.
The Washington Post reported that police, confident in their unproven facial recognition technology, are skipping investigative steps and ignoring standards. At least 8 Americans have been wrongfully arrested due to this so far.
After 60 years of dormancy, the world’s first chatbot ELIZA has been revived by AI historians and computer scientists on its original operating system, bringing a piece of computing history back to life.
Phind.com is an Ai chatbot designed for developers.
AI is good at coding and rewriting content but some researchers have found it is not good at history.
Search Engine Land has identified 4 ways to use AI in copywriting.
AI For Good
New Insights in Diabetes Management
Today, about 38 million Americans (11.6% of the population) live with diabetes, with an additional 97.6 million (38% of the population) identified as prediabetic. Traditionally, diabetes has been categorized into two main types: Type 1, which typically appears in childhood, and Type 2, which can develop later in life. However, Type 2 diabetes encompasses a wide range of variations, prompting Stanford researchers to create an AI-based algorithm. This innovative tool analyzes data from continuous blood-glucose monitors to identify hyper-specific Type 2 sub-categories.
This subclassification is important for assessing an individual's risk of developing related conditions, such as liver, kidney, and eye complications, and for evaluating the effectiveness of various drugs and treatment plans. In a study involving 54 participants, the algorithm successfully identified metabolic subtypes, including insulin resistance and beta-cell deficiency, with 90% accuracy, outperforming traditional metabolic tests.
Why it matters: "It's a tool that people can use to take preventative measures," says Dr. Michael Snyder, a professor of genetics and co-leader of the study. This means if levels indicate a prediabetes warning, lifestyle adjustments in diet or exercise can be made proactively.
Dr. Tracey McLaughlin, a professor of endocrinology, emphasizes the importance of this information, noting that even without developing diabetes, insulin resistance poses risks for other health conditions like heart disease and fatty liver disease.
Tools That Work
Cudo Compute is a cloud-based service provider that offers high-performance computing, AI, and deep learning solutions.
Dubsado is great for contract writing and project management.
Folk is the number one AI powered CRM tool.
N8N is the most powerful automation tool
Unlock AI's Potential through Collaborative Approaches
Asking AI to “act like a team of PhD researchers” seems to significantly enhance its output. This method taps into the potential of large language models by combining structured collaboration, advanced reasoning, and psychological techniques. By adopting expert personas, AI doesn't just simulate knowledge; it creates a dynamic, collaborative problem-solving system in its responses.
This collaborative approach suggests that leveraging multiple expert perspectives can significantly boost the accuracy and quality of results. For instance, simply reviewing code for errors can reduce mistakes, but simulating a review by a group of top PhD researchers improves the output even further. This mirrors the ReAct framework in agentic systems, combining reasoning with action and reflection, allowing AI to self-correct, refine logic, and produce more robust outcomes.
Recent performance analyses show a dramatic improvement in task completion accuracy, with an almost 85% increase, faster response times, and moderate token consumption. While this approach is more verbose and costly, Microsoft’s research indicates a performance boost of over 10% when emotional and professional dynamics are integrated into prompts. These improvements demonstrate that collaborative structures enhance accuracy and optimize efficiency, marking a significant leap in capability.
However, these methods are not without drawbacks. Effectiveness depends on task complexity, fidelity of role representations, and quality of example data. Overly complex role assignments can lead to diminishing returns, potentially over-analyzing aspects requiring little analysis.
Looking to the future, AI and agent-centric systems will likely be defined by collaborative architectures rather than single-agent systems. Emerging collaboration styles include swarm systems, where decentralized agents share real-time updates; hierarchical teams with specialized roles; and hybrid ensembles integrating distinct AI and human agents. These systems thrive on constant communication and iterative improvement, leading to exponential increases in both speed and quality of output.
In this next wave of development, collaborative AI will evolve from a powerful tool into a networked intelligence, significantly enhancing the ability to think, solve, and create.
How AI Tutoring in Nigeria Achieves Two-Year Learning Gains
The World Bank's research, as highlighted by Ethan Mollick's post on X, presents a compelling case for the transformative potential of AI in education. The study focused on a controlled trial where students in Nigeria received after-school tutoring from GPT-4, an advanced AI language model developed by OpenAI. Over the course of six weeks, these students experienced extraordinary learning gains equivalent to about two years of traditional education. This outcome not only surpassed the typical results of 80% of other educational interventions but also demonstrated the ability of AI to significantly enhance learning across diverse student groups.
One of the most remarkable aspects of the trial was its inclusive impact. While all students benefited from the AI tutoring, the gains were particularly pronounced among girls who initially lagged behind their peers. This highlights AI's potential to bridge educational gaps and promote equity in learning opportunities. The AI tutor's ability to provide personalized and patient instruction allowed students to engage with the material at their own pace, fostering a deeper understanding and retention of knowledge.
The assertion that AI will eventually disrupt every job underscores the need for individuals and organizations to adapt to this technological evolution. In the realm of education, AI's attributes as a patient, always-available, and creative tutor make it an invaluable resource for learners seeking to enhance their understanding of complex subjects.
AI's capacity to generate numerous examples and explanations tailored to individual learning styles is a game-changer in educational contexts. This adaptability ensures that students can internalize key insights and concepts effectively. Additionally, AI's tireless nature means it can provide assistance at any time, accommodating different learning schedules and preferences.
However, while AI offers immense potential, it is crucial to approach its use with a discerning eye. Users should always verify the information provided by AI, as these models, despite their sophistication, can occasionally make errors. This underlines the importance of maintaining a critical perspective and cross-referencing AI-generated insights with reliable sources.
In my personal experience, I utilize AI daily to enhance my understanding of various topics. This application accounts for a significant portion of my interactions with AI models, which have proven to be incredibly effective in facilitating deeper comprehension. Nonetheless, it is important to remain vigilant and ensure the accuracy of the information received from AI.
In conclusion, the World Bank's study on AI tutoring in Nigeria offers a glimpse into the future of education, where AI plays a pivotal role in democratizing learning and fostering equitable access to quality education. The impressive learning gains achieved through AI tutoring underscore its potential to revolutionize educational practices and outcomes. As AI continues to evolve, its integration into education and other sectors will undoubtedly shape the future of work and learning, making it imperative for individuals and institutions to embrace and adapt to these technological advancements.
