AI for Education

Today is National AI Literacy Day and I'm excited to share that I’ll be speaking on a panel for Congressional staff this afternoon and hosting a special free virtual workshop you can still sign up for. It's incredible to see how an idea I shared with colleagues is now a day of AI literacy learning with events around the country. The AI for Education team will be leading our own session on “Building AI Literacy in Your School or District” at 11am EST. It’s not too late to register and even if you can’t make it, we’ll be sharing the recording and resources with all the registrants! Link in the comments to sign up! #ailiteracyday #ailiteracy #GenAI

This week OpenAI announced their new GPT4o image generator, calling it a "game changer for educators." They showed some incredible examples, including a correctly-labeled?cross section of a cell. The former biology teacher in me could not help but get excited, and the team here at AI for Education loves an image generator, so we all dug into explore further and found the techology is not quite there yet. There's no doubt that the new 4o image generator is a major improvement on the Dall-E integration. Being able to directly prompt GPT4o to create the image instead of the LLM prompting an image generator means a lot more control over the outputs. We also love the improvements on text rendering and the consistent characters. And Mandy DePriest on our team got it to correctly render the lyrics to Abba's Dancing Queen, her personal?benchmark. But when we used the same prompt from the original post to try and duplicate the labeled cell, we repeatedly came up short. Despite trying various prompting strategies, it could never accurately create a diagram of a cell and label the parts. For comparison's sake?we ran the same prompt through multiple generators, with poor results. While image generators are improving quickly, it's clear they still have a lot of room for improvement. But sometimes the mistakes are the best teachable moments. As Hannah Ketteman on our team shared, one of her favorite things is to produce an image like?these and then have students find and correct the errors - a great way to assess students while building their AI literacy. What do you think of 4o's new image capabilities? What are your favorite applications and those you are hoping improve? #GenAI #AI #ailiteracy

As part of AI for Education's recent webinar on AI ethics, Corey Layne Crouch and I explored a number of tools during our discussion of deepfakes including a hyper-realistic conversation with "Miles" - a voice AI demo from Sesame AI. As a Voice AI system designed to feel as "human" as possible, it's a great demonstration of what makes these technologies both fascinating and concerning from an ethical standpoint. Miles sounds incredibly real, and quickly adapted to the direction of our conversation and often adjusted it's responses to agree with us. When I mentioned a previous AI conversation in a workshop where the female counterpoint "Maya" claimed to be human, Miles highlighted this as the exact ethical problem we need to address: "People might start relying too heavily on us, forgetting that we're not human. It's easy to get comfortable with a really convincing voice." Miles certainly hit the nail on the head when it concluded that "as AI gets more sophisticated, these lines are going to get blurrier and blurrier." The experience was a perfect example of why we need AI literacy and robust conversations about transparency and disclosure of use, significant guardrails, and responsible development as these deepfake technologies become increasingly convincing. Take a listen to our conversation and try Sesame Voice AI for yourself through the link in the comments. #AIEthics #Deepfakes #VoiceAI #AILiteracy

We’re happy to welcome Mandy DePriest as a panelist in our AI in Education: Roundtable Series! ?? Join us for two insightful sessions on AI’s role in enhancing education: ??? March 26: How AI Can Help Improve the Quality of Education ??? April 2: How AI Can Help Improve the Delivery of Education Moderated by Scott Fisher - don’t miss out! ?? Register here: https://lnkd.in/eympppEw

A new study by METR (Model Evaluation & Threat Research) shows GenAI systems are getting exponentially better at completing longer tasks, doubling their capabilities every 7 months. Key Findings From the Report: ? Researchers introduced "50%-task-completion time horizon" as a new metric - the time humans typically take to complete tasks that AI models can complete with 50% success rate ? Current frontier AI models (like Claude 3.7 Sonnet) have a time horizon of approximately 50 minutes ? AI time horizon has been doubling approximately every seven months since 2019, with acceleration in 2024 ? Improvements appear driven by greater reliability, ability to adapt to mistakes, and enhanced logical reasoning and tool use capabilities ? Testing across 170 tasks of varying difficulty from software engineering, cybersecurity, and ML research showed consistent exponential improvement ? There's a clear pattern: models have almost 100% success on tasks taking humans <4 minutes, but <10% success on tasks taking humans >4 hours ? Analysis suggests that if trends continue, within 5 years AI systems could be capable of automating many software tasks that currently take humans days or weeks to complete This research provides a helpful quantitative framework for measuring AI capabilities against human performance. It also provides even more evidence of why AI literacy and training are imperative now, as the technology's capacity continues to grow exponentially. AI for Education #GenAI #AIliteracy #AI

Yesterday, we had over 400 people join us for AI for Education's webinar on "Navigating AI Ethics in Education: From Deepfakes to Academic Integrity." Corey Layne Crouch and I co-hosted a thought-provoking discussion on key ethical considerations for educators. We discussed and provided strategies/resources for: ? Rethinking academic integrity ? Dealing with the rising sophistication of deepfakes (images, videos, and voice) ? Discussing the environmental impact of AI? ? Shifting from reactive to proactive strategies when dealing with ethical concerns of AI One of the highlights of the session was demoing Sesame's Voice AI, which has been designed to be as believable as possible. This definitely blew some of the audiences minds! We suggest watching if you are interested in learning more about these critical issues and diving into strategies for ethical AI use that can empower both educators and students to navigate this complex technology with confidence. Link in the comments to watch the recording and checkout the resources mentioned during the session. And look out for a Part 2 in late April! #ailteracy #ethicalAI #GenAI #AI