|
| Together with |
 |
|
|
|
|
|
Hi there, this is your daily ☕️ Techpresso.
|
|
|
In today's Techpresso: |
|
🧠 Google unveils Gemini 2.5
🎨 OpenAI unveils new image generator for ChatGPT
🔮 Microsoft introduces deep research and analysis tools for Copilot
🎵 Anthropic wins against music publishers
🔬 Google quantum chief predicts major breakthrough within 5 years
❌ Apple barred from Google antitrust trial
🎁 + 14 other news you might like
🔮 + 2 handpicked research papers and tools
|
|
FROM OUR PARTNER
BOXABL officially reserved NASDAQ stock ticker symbol BXBL...
BOXABL is the home construction company bringing assembly line automation to the home industry. With their patented technology and 53 patent filings, BOXABL believes they have the potential to disrupt a massive and outdated trillion dollar building construction market. And they just reserved the ticker symbol BXBL on Nasdaq*!
Most houses take seven months to complete. BOXABL can put one out of the assembly line every four hours, including electrical, HVAC, and plumbing!
Currently, shares are being offered at $0.80 per share. With a $1,000 minimum investment, you can join 40,000 investors to help solve the housing crisis.
Time is running out. The closing date 3/29 is the last day to invest in BOXABL on StartEngine.
Invest Now!
| 🧠 Google unveils Gemini 2.5 LINK
|
- Google has launched Gemini 2.5 Pro Experimental, described as their "most intelligent" AI model yet, featuring enhanced reasoning capabilities, multimodality, and a massive context window that outperforms competing models in various benchmarks.
- The new model incorporates built-in reasoning that essentially fact-checks itself during output generation, significantly improving its performance, particularly in "agentic" coding tasks where it can create fully functional video games from a single prompt.
- Gemini 2.5 Pro boasts an impressive 1 million token context window, allowing it to process multiple lengthy books in a single prompt, while also achieving record-breaking scores in complex benchmarks like Humanity's Last Exam.
| 🎨 OpenAI unveils new image generator for ChatGPT LINK
|
- OpenAI has replaced DALL-E with a new native image generator directly integrated into ChatGPT, promising more consistent results and fewer content restrictions for all users.
- The upgraded system processes text and images together, allowing it to handle up to 20 different objects while maintaining correct spatial relationships and showing strength with unconventional concepts.
- Users can refine images through natural conversation with the AI maintaining context across multiple exchanges, though the system still struggles with accurately rendering text and certain complex scenes.
| 🔮 Microsoft introduces deep research and analysis tools for Copilot LINK
|
- Microsoft is launching two new "reasoning agents" for Copilot called Researcher and Analyst, designed to analyze extensive work data across emails, meetings, chats, documents, and external sources.
- The Researcher tool combines OpenAI's deep research model with Copilot's orchestration capabilities to conduct complex research tasks, while creating comprehensive reports and strategies using both internal and web data.
- Built on OpenAI's o3-mini reasoning model, the Analyst tool functions like a data scientist, processing multiple spreadsheets to forecast financial outcomes and visualize customer patterns, with both tools becoming available in April.
| 🎵 Anthropic wins against music publishers LINK
|
- A California judge denied a motion filed by major music publishers that would have prevented Anthropic from using song lyrics, stating the requested injunction was "poorly defined" and could impose "unforeseeable costs" on the company.
- Music publishers including Universal Music Group and Concord had alleged in their October 2023 lawsuit that Anthropic "unlawfully copies and disseminates vast amounts of copyrights works" while building its AI models.
- The judge ruled that the publishers failed to demonstrate reputational harm or market value diminishment from Anthropic's use of their work, highlighting the ongoing legal battles between AI companies and copyright holders.
| 🔬 Google quantum chief predicts major breakthrough within 5 years LINK
|
- Google Quantum AI's director Julian Kelly predicts quantum computers will achieve a practical breakthrough within five years, enabling calculations impossible for traditional computers to solve.
- Currently, Google's most advanced quantum computer has 105 qubits, while experts believe one million or more will be needed for useful applications in physics simulations and potentially generating novel data.
- The quantum computing field has gained attention following Google's error correction breakthrough in December, though Nvidia CEO Jensen Huang has expressed skepticism about the technology's near-term practicality.
| ❌ Apple barred from Google antitrust trial LINK
|
- Apple has been blocked from participating directly in an antitrust hearing involving its $20 billion annual search deal with Google because its lawyers were too slow to file their intervention request.
- The Department of Justice's case against Google determined the tech giant violated antitrust laws through deals making it the default search engine on browsers, including Safari on Apple devices.
- While unable to present live testimony or cross-examine witnesses, Apple can still submit written testimony and file friend-of-the-court briefs to protect its interests in the proceedings.
|
|
FROM OUR PARTNER
AI is no longer the future. It's here, and it's revolutionizing every industry & the way we work.
Introducing a 2-Day Free AI Bootcamp by Outskill which comes with 16 hours of intensive training on AI frameworks & use-cases that can make you an AI expert.
Inside the AI Bootcamp, you will learn:
👉 AI tools to automate repetitive tasks and free up time for high-value work.
👉 Smarter decision-making with Generative AI, Neural Networks, and LLMs.
👉 AI-generated images and videos to speed up content creation and marketing.
Join in now, (we have limited free seats! 🚨)
|
|
|
|
Other news & articles you might like
- The TikTok ban is losing support among Americans, study says LINK
- US adds dozens of Chinese entities to export blacklist, including Inspur units LINK
- Apple's CEO met with 'the next generation of developers' in DeepSeek's hometown LINK
- Lisa Su says Radeon RX 9000 series is AMD's most successful GPU launch ever LINK
- OpenAI's top models crash from 75% to just 4% on challenging new ARC-AGI-2 test LINK
- Qualcomm cries anti-competition to regulators in continued Arm fall out LINK
- Devs say AI crawlers dominate traffic, forcing blocks on entire countries LINK
- This company is using AI to give people American-sounding accents LINK
- Apple is about to spend $1 billion on NVIDIA servers for AI: analyst LINK
- China built hundreds of AI data centers to catch the AI boom. Now many stand unused. LINK
- Amazon is testing shopping, health assistants as it pushes deeper into generative AI LINK
- Signal defends itself after U.S. military officials leak classified plans by mistake on group chat LINK
- Some Google Pixel owners reporting failed alarm clocks LINK
- GameStop hints at future Bitcoin purchases following board approval LINK
|
|
Latest research and tools
|
Optimizing ML training with metagradient descent: this paper presents a method that uses gradients of gradients to automatically tune hyperparameters during neural network training, improving efficiency and performance. LINK
|
|
SplitQuantV2: Enhancing Low-Bit Quantization of LLMs Without GPUs: researchers present a technique that improves low-bit quantization of large language models without requiring GPU resources. LINK
|
|
|
|
|
Want to get the latest news differently? |
|
|
|
|
