Ashish Bansal (@ash_bans) header
Ashish Bansal (@ash_bans) profile picture

Machine Learning makes predictions. Artificial Intelligence takes decisions. © CEO @StarSparkAI, ex-Google/Twitch/Twitter. https://teapawt.net/t.co/xoqOzByLne / banaras.eth

San Francisco, CA
Joined April 2012
On @edufuturists this week: can we personalise learning at scale? We talked Socratic AI tutors at @StarsparkAI, the multiple-choice guessing problem, and why math gaps compound. youtu.be/3BqWHQE7kfw
2
268
Thanks for recognizing our effort to use AI in purpose built education systems! @StarsparkAI
Now that August has arrived (man, that was fast!), the back-to-school season vibes are stirring and kids have to prep themselves for another year of in-person studies. With the continued evolution of AI tools, teachers and students alike have been using them to assist them in their everyday learning. While it’s disappointing to hear stories about students producing their work entirely with AI, it’s equally satisfying to see AI tools such as Gemini Notebooks, Perplexity, and ChatGPT offer educational-themed features that don’t encourage cheating. Speaking of AI tools built to educate students, one of the most prominent ones is StarSpark, an AI-powered math and science platform that’s designed to teach rather than simply give out answers. It primarily covers K-12 and AP coursework while adapting to a student’s specific grade level, learning style and the knowledge they’ve already acquired. And for parents, StarSpark is just as beneficial for them—the AI educational tool provides them with accessible, on-demand support that helps them grasp the way math and science are now taught.
1
2
173
Ashish Bansal retweeted
StarSpark is in @Forbes 🎉 We focus on teaching the process, not just the answer, and this piece explains why that's the whole game. Read the article here: forbes.com/sites/rayravaglia… #aiineducation #aitutor #forbesfeature #starsparkai #k12learning #edtech
1
1
2
67
A few years ago my 5th grader signed up for an advanced online math course and was acing it. I asked how. He said: "I know all the questions and answers, so I just memorized them." The system measured right answers. It had no idea if he understood anything. That's why @rohit_joshi and I built @StarsparkAI .
2
1
2
245
The hard part: #LLMs are trained to be agreeable. But learning doesn't always come from agreement. Sometimes a student needs to be slowed down or told to try again. Getting an #AI to push back politely but firmly, instead of caving, is a real design challenge.
1
1
12
This is an amazing recognition of our high impact work in making quality affordable #AI tutoring accessible to all! #EdTech #AIinEducation #Math
Exciting news for StarSpark AI! We have been awarded a grant to use AWS AI and cloud technology through the AWS Education Equity Initiative, supporting our mission to make high quality, personalized learning accessible for every student. Thank you @awscloud! #AWSImpact
54
Ashish Bansal retweeted
Big news! StarSpark was featured on KFDX Texas News to break down the infamous "Summer Slide" and how it impacts our students 📉 Did you know that kids lose an average of 2.6 months of math skills over the summer break? It is just like a muscle; if you do not use it, you lose it! Our Co-founder & CEO, Ashish Bansal, dropped some incredible wisdom on air about how we can build small, 30-minute daily habits to keep those math batteries charged all summer long. 🧠✨ This is a huge milestone for us, and it is only the beginning. We have some massive announcements and game-changing updates coming very soon, so make sure you are following along! 🚀 Watch the full segment here: texomashomepage.com/news/loc… #StarSparkAI #EdTech #InTheNews #SummerSlide #StartupLife #MathIsALifeSkill #KFDXNews
2
2
58
"I just took the same screenshot you showed me, on my machine, with my code applied. It works. Yours doesn't." @AnthropicAI 's Claude Opus 4.7 xhigh after being unable to fix a simple tailwind CSS issue it had introduced. @OpenAI 's Codex: "The new theme classes trigger malformed generated CSS because tailwind.config.js has broken custom font-size fallbacks for...." on GPT5.5 High. Codex 1 - Claude 0
1
114
Ashish Bansal retweeted
Yay, finally! Introducing Vision Banana🍌 from @GoogleDeepMind, our unified model that outperforms SoTA specialist models on various vision tasks! By treating 2D/3D vision tasks as image generation, we unlock a new foundation for CV. Project page: vision-banana.github.io (1/5)
58
306
48
2,199
291,867
Ashish Bansal retweeted
Math education is evolving quickly, and parents are doing their best to keep up with a shifting landscape and the growing influence of AI in learning. To support parents, we partnered with @OpenStax to host a free webinar led by Meg Knapik, Educational Advisor and former Superintendent of Curriculum and Instruction in Illinois, and Ashish Bansal, CEO and Co-Founder of StarSpark.AI. They will break down how kids learn math best in today’s education environment and how families can support that learning at home. In this 45-minute session, we will cover: • The difference between school curriculum and State Standards • How Singapore and U.S. math approaches shape student success • How AI tutoring tools like StarSpark build mastery, confidence, and consistent practice 📅 December 4 at 4 PM CST / 5 PM EST 🔗 Register at the link in our bio #parentwebinar #matheducation #aiinlearning #mathhelp #mathematics #education
1
1
3
175
Ashish Bansal retweeted
RIP fine-tuning ☠️ This new Stanford paper just killed it. It’s called 'Agentic Context Engineering (ACE)' and it proves you can make models smarter without touching a single weight. Instead of retraining, ACE evolves the context itself. The model writes, reflects, and edits its own prompt over and over until it becomes a self-improving system. Think of it like the model keeping a growing notebook of what works. Each failure becomes a strategy. Each success becomes a rule. The results are absurd: +10.6% better than GPT-4–powered agents on AppWorld. +8.6% on finance reasoning. 86.9% lower cost and latency. No labels. Just feedback. Everyone’s been obsessed with “short, clean” prompts. ACE flips that. It builds long, detailed evolving playbooks that never forget. And it works because LLMs don’t want simplicity, they want *context density. If this scales, the next generation of AI won’t be “fine-tuned.” It’ll be self-tuned. We’re entering the era of living prompts.
232
1,126
183
7,659
717,204
Empty Microsoft Word files and @AnthropicAI Claude Code new sessions are about the same size. @alexalbert__ if 87K of 200k is taken with context, it severely limits ability to do stuff. I have had to remove tools and context files.
1
146
This is super important topic that causes a lot of confusion for parents and students. Meg has an excellent article explaining the difference!
There is often confusion between standards and curriculum, but they are not the same thing...think about it like the destination vs. the journey to get there! starspark.ai/education-stand… @StarsparkAI
2
99
Ashish Bansal retweeted
I’ve been working on something new: 📚 Build a Reasoning Model (From Scratch). The first chapters just went live! (The book will cover topics from inference-time scaling to reinforcement learning)
106
343
23
2,927
160,210
Just experienced the legendary Taj hospitality @TajCampton when they heard my friend and I talking about plans for my birthday and they bought cake! Thank you!
2
139