Gemini 2.0 is an experimental chatbot model from Google that uses your Google search history as a context to personalize the conversation. This is a unique and somewhat controversial feature, as it allows the chatbot to tailor its responses based on your personal search data.

The model is accessible through the Gemini.com website, where you can try it out for free by logging in with your Google account. Once logged in, you can switch to the "personalization" model, which will use your search history to inform the chatbot's responses.

The results can be quite insightful, as the chatbot is able to pick up on your interests and tailor the conversation accordingly. However, this feature also raises privacy concerns, as it gives the chatbot access to your personal search data.

While this level of personalization can be useful, it's important to be aware of the potential implications. Users who value their privacy may be hesitant to use this feature, as it could expose sensitive information. Additionally, the chatbot's responses may not always be appropriate or consistent, as it is relying on your search history rather than a curated set of instructions.

Overall, Gemini 2.0's use of search history personalization is a bold and innovative move by Google, but it comes with its own set of considerations and potential drawbacks. Users should carefully weigh the benefits and risks before deciding whether to utilize this feature.

Gemini, the AI platform, has recently introduced several updates to its offerings, including a new canvas feature that allows users to run Python code, similar to features available on other large language models (LLMs) like Claude and GPT.

One notable update is the addition of the canvas feature, which was previously only available on the Flash model. This feature enables users to execute Python code directly within the Gemini interface, providing a more interactive and hands-on experience.

Additionally, Gemini has improved its Deep Research feature, which now aligns more closely with the capabilities of ChatGPT and other leading LLMs. While it may still be slightly inferior to ChatGPT, the updates have helped Gemini catch up to its competitors in terms of functionality.

Another unique feature introduced by Gemini is the audio review option. When users run a Deep Research task, they can now generate a concise audio overview of the research, similar to a podcast-like conversation. This feature can be particularly useful for quickly summarizing and digesting lengthy research documents.

Overall, Gemini's recent updates demonstrate the platform's efforts to keep pace with the rapidly evolving AI landscape. By incorporating features found in other popular LLMs, Gemini is aiming to provide users with a more comprehensive and competitive set of tools for their AI-powered needs.

If you want to try out the new Chinese AI models Ernie 4.5 and Ernie X1, you'll encounter a few challenges. The website doesn't offer translations, so you'll need to use Google Translate to understand the content. Additionally, you'll need a Chinese phone number to actually try out the models.

While these models are capable benchmarks, the main selling point is the price. Ernie 4.5 is a potential GPT-4.5 competitor, while Ernie X1 is a deep-seek or GPT-3 mini competitor. The Ernie X1 model is claimed to be half the price of GPT-R1, which was already a fraction of the price of Western models.

Compared to GPT-4.5, the Ernie 4.5 model scores on par, but the input tokens are less than 1% of the price. At 55 cents per 1 million input tokens and 220 for 1 million output tokens, it's roughly 1/5th of the price of GPT-4.5.

However, without being able to test the models directly, it's difficult to say whether they are truly better than GPT-4.5. The outputs seen so far suggest they are capable, but not necessarily better than the GPT-4.5 model. The essence of this release is that new, highly capable Chinese models are emerging, and the price trend continues to be significantly lower than their Western counterparts.

This week, the online entrepreneur and "Builder of Curious Things" Christopher Columbus is organizing the 2023 Vibe Coding Game Jam. The event is a showcase of creativity and the power of AI-assisted game development.

The premise is simple - participants can use tools like Cursor and AI models like Sonet 3.7 to "vibe code" and rapidly build games. This approach has already proven successful, with Columbus himself creating a Flight Simulator game that attracted over 320,000 players and generated $887,000 in revenue in just one month.

The Vibe Coding Game Jam has attracted the attention of notable figures like AI researcher Andre Karpathy, who will be serving as one of the judges. This event presents a unique opportunity for both seasoned and aspiring game developers to explore the potential of AI-powered game creation.

Whether you're an experienced programmer or a creative individual looking to dip your toes into game development, the Vibe Coding Game Jam is worth checking out. With just a few days left, you can follow the event on Twitter and see the incredible games that emerge from this AI-fueled creative process.

This website allows you to test and compare the performance of different language models in building structures within the game Minecraft. The challenge is designed to assess the AI's ability to understand and execute complex spatial and construction tasks.

On the site, you can view different structures created by various models and vote on which one looks more like the intended object or design. The leaderboard shows the top-performing models, with Claude 3.7 and Sonet currently leading the pack.

This benchmark is a unique and engaging way to evaluate the capabilities of language models in a visual, interactive environment. By seeing the models' actual creations, you can get a better sense of their reasoning and planning abilities, rather than relying solely on abstract test scores.

If you're interested in AI and its applications, I recommend bookmarking this page and trying out the benchmark yourself. It's a fun and informative way to explore the current state of language model performance in a creative, hands-on task.

Roblox, the massively popular game platform with over 380 million monthly active users, has recently integrated a new generative AI system called Cube. This system allows users to create 3D models from text or voice prompts, opening up a new level of creativity for the game's 2.5 million developers.

Cube's integration into Roblox is significant because it targets a vast audience, with nearly half of the platform's users being under the age of 13. This means that the next generation of creators will have access to powerful AI tools to bring their ideas to life, without the need for extensive 3D modeling skills.

The impact of Cube goes beyond just the Roblox platform. As these AI-powered tools become more accessible, they have the potential to democratize 3D content creation, empowering a new wave of young creators to express their creativity in ways that were previously reserved for specialized artists and designers.

This development highlights the growing influence of generative AI in the gaming and entertainment industries. As these technologies continue to evolve, we can expect to see more innovative applications that blur the lines between consumption and creation, shaping the experiences of the next generation of digital natives.