GenAI Weekly — Edition 11
Your Weekly Dose of Gen AI: News, Trends, and Breakthroughs
Stay at the forefront of the Gen AI revolution with Gen AI Weekly! Each week, we curate the most noteworthy news, insights, and breakthroughs in the field, equipping you with the knowledge you need to stay ahead of the curve.
Rabbit R1, a thing that should just be an app, actually is just an Android app
When startups like Humane and Rabbit first unveiled their AI-powered gadgets a couple of months ago, early backers were hopeful that these AI companions would usher in a new era of wearable AI. Unfortunately for them, both products launched half-baked, leaving tech reviewers with no choice but to harshly criticize them.
The Rabbit R1, for example, does almost nothing that your Android phone can’t do. In our hands-on, we noted that the gadget lets you do things like talk to a large language model (LLM) to get answers to questions, take a photo of an object to get info about it, play music from Spotify, hail a ride from Uber, or order food from Doordash, but that’s basically it. If everything an AI gadget like the Rabbit R1 can do can be replicated by an Android app, then why aren’t these companies simply releasing an app instead of hardware that costs hundreds of dollars, requires a separate mobile data plan to be useful, and has terrible battery life? It turns out that’s exactly what Rabbit has done…sort of.
See, it turns out that the Rabbit R1 seems to run Android under the hood and the entire interface users interact with is powered by a single Android app. A tipster shared the Rabbit R1’s launcher APK with us, and with a bit of tinkering, we managed to install it on an Android phone, specifically a Pixel 6a.
See also: Analysis of the Rabbit R1 APK
Microsoft introduces GitHub Copilot Workspace
In the past two years, generative AI has foundationally changed the developer landscape largely as a tool embedded inside the developer environment. In 2022, we launched GitHub Copilot as an autocomplete pair programmer in the editor, boosting developer productivity by up to 55%. Copilot is now the most widely adopted AI developer tool. In 2023, we released GitHub Copilot Chat—unlocking the power of natural language in coding, debugging, and testing—allowing developers to converse with their code in real time.
After sharing an early glimpse at GitHub Universe last year, today, we are reimagining the nature of the developer experience itself with the technical preview of GitHub Copilot Workspace: the Copilot-native developer environment. Within Copilot Workspace, developers can now brainstorm, plan, build, test, and run code in natural language. This new task-centric experience leverages different Copilot-powered agents from start to finish, while giving developers full control over every step of the process.
[…]
For developers, the greatest barrier to entry is almost always at the beginning. Think of how often you hit a wall in the first steps of a big project, feature request, or even bug report, simply because you don’t know how to get started. GitHub Copilot Workspace meets developers right at the origin: a GitHub Repository or a GitHub Issue. By leveraging Copilot agents as a second brain, developers will have AI assistance from the very beginning of an idea.
[…]
From there, Copilot Workspace offers a step-by-step plan to solve the issue based on its deep understanding of the codebase, issue replies, and more. It gives you everything you need to validate the plan, and test the code, in one streamlined list in natural language.
Friends From the Old Neighborhood Turn Rivals in Big Tech’s A.I. Race
领英推荐
Their path from London to the executive suites of Big Tech is one of the most unusual — and personal — stories in an industry full of colorful personalities and cutting rivalries. In 2010, they were two of the three founders of DeepMind, a seminal A.I. research lab that was supposed to prevent the very thing they are now deeply involved in: an escalating race by profit-driven companies to build and deploy A.I.
Their paths diverged after a clash at DeepMind, which Google acquired for $650 million in 2014. When the A.I. race kicked off in late 2022 with the arrival of the ChatGPT online chatbot, Google put Dr. Hassabis in charge of its A.I. research. Mr. Suleyman took a rockier route — founding another A.I. start-up, Inflection AI, that struggled to gain traction before Microsoft unexpectedly hired him and most of his employees.
“We’ve always seen the world differently, but we’ve been aligned in believing that this is going to be the next great transition in technology,” Mr. Suleyman said of his old family friend in an interview. “It is always a friendly and respectful rivalry.”
New wave, new leadership.
Secret Llama: Run models directly in your browser with WebGPU
Secret Llama is a free and fully private chatbot. Unlike ChatGPT, the models available here run entirely within your browser which means:
Note: the first message can take a while to process because the model needs to be fully downloaded to your computer. But on future visits to this website, the model will load quickly from the local storage on your computer.
Supported browsers: Chrome, Edge (GPU required; Mobile not recommended)
While projects like Ollama let you run LLMs on your computer, SecretLlama’s approach is to run LLMs right in your browser. Of course LLMs need access to a GPU for acceptable performance. So, how does that happen in a browser? That’s where WebGPU comes in. Of course, only some of the latest browsers support this standard now, but one can safely assume that WebGPU support should be pretty standard in the coming years.
AI Copilots Are Changing How Coding Is Taught
“Students are early adopters and have been actively testing these tools,” says Johnny Chang, a teaching assistant at Stanford University pursuing a master’s degree in computer science. He also founded the AI x Education conference in 2023, a virtual gathering of students and educators to discuss the impact of AI on education.
So as not to be left behind, educators are also experimenting with generative AI. But they’re grappling with techniques to adopt the technology while still ensuring students learn the foundations of computer science.
“It’s a difficult balancing act,” says Ooi Wei Tsang, an associate professor in the School of Computing at the National University of Singapore. “Given that large language models are evolving rapidly, we are still learning how to do this.”
[…]
As a result, educators are modifying their teaching strategies. “I used to have this singular focus on students writing code that they submit, and then I run test cases on the code to determine what their grade is,” says Daniel Zingaro, an associate professor of computer science at the University of Toronto Mississauga. “This is such a narrow view of what it means to be a software engineer, and I just felt that with generative AI, I’ve managed to overcome that restrictive view.”
Zingaro, who coauthored a book on AI-assisted Python programming with Porter, now has his students work in groups and submit a video explaining how their code works. Through these walk-throughs, he gets a sense of how students use AI to generate code, what they struggle with, and how they approach design, testing, and teamwork.
Programming has changed forever, whether you like it or not. It’s a good idea to figure the needed changes in how to approach it academically.
For the extra curious