- Evolving AI Insights
- Posts
- š Gemini 1.5 Pro is NEXT-LEVEL
š Gemini 1.5 Pro is NEXT-LEVEL
Also: SoftBank's AI Chip Ambition
Welcome, AI enthusiasts
Google was ready to dominate the headlines with their Gemini 1.5 Pro announcement, but OpenAIās Sora text-to-video model quickly overshadowed it. However, as the weekend progressed, more and more news emerged suggesting that the Gemini 1.5 Pro might just be the groundbreaking LLM weāve been anticipating. Softbankās CEO announced plans to raise a substantial amount of money for an AI chip company, aiming to rival Sam Altmanās venture, and Magic AI secured $117 million to develop an AI software engineer. Letās dive in!
In todayās insights:
Gemini 1.5 Pro is NEXT-LEVEL
SoftBank's AI Chip Ambition
Magic AI Secures $117 Million to Build an AI Software Engineer
Read time: 5 minutes
šļø LATEST DEVELOPMENTS
GOOGLE
š Gemini 1.5 Pro is NEXT-LEVEL
Evolving AI: Google has achieved the most awaiting comeback with Gemini 1.5 Pro.
Key Points:
Gemini 1.5 Pro has a staggering 99.7% accuracy rate.
Can process millions of tokens and find the right answer in 1-2 minutes.
A monumental step forward in Generative AI.
Details:
Google showcased the impressive abilities of Gemini 1.5 Pro through a test known as 'Needle in a Haystack'. This test evaluates the precision with which a large language model (LLM) can sift through and remember details from extensive data. Surprisingly, Google reports that Gemini 1.5 Pro can recall information with an astonishing 99.7% accuracy from contexts including 1 million tokens of text, 22 hours of audio, and 3 hours of video. This performance has been verified by individuals who have had the opportunity to test the model themselves. To understand the significance of this, consider that the most sophisticated LLMs available to the public until now have managed contexts ranging from 32,000 to 200,000 tokens. Achieving 99.7% accuracy with 1 million tokens marks an extraordinary breakthrough, significantly advancing the capabilities of AI systems in understanding and reasoning.
Google points out that Gemini 1.5 Pro is at the forefront of models based on their own research, building on the foundational Transformer paper and their groundbreaking work in developing mixture-of-experts architectures.
This indicates that Gemini 1.5 Pro is the culmination of years of LLM research and development, rather than a sudden innovation. Google's expertise in creating the foundational technologies for Gemini has been crucial in bringing this groundbreaking model to practical use.
Our Thoughts:
Gemini 1.5 Pro signifies a monumental stride forward, setting new benchmarks for the LLM field. The research community and companies such as Anthropic, Meta, and DeepMind are now challenged to meet or surpass these new standards in context length, multi-modality, precision, and practical application.
The quest for models inching towards Artificial General Intelligence (AGI) is speeding up, fueled by milestones like Gemini 1.5 Pro. We are keen to see the next breakthroughs as researchers relentlessly explore new horizons in AI.
Watch the demo below by Google to see how quickly Gemini 1.5 Pro can sift through context and provide the answers you're looking for.
Gemini 1.5 Pro can understand tasks and questions across different modalities because of its long context understanding. When given a 44-minute Buster Keaton film, it's able to find small details in the film and understand plot points. #GeminiAI
ā Google (@Google)
7:20 PM ā¢ Feb 15, 2024
SOFT BANK
š SoftBank's AI Chip Ambition
Evolving AI: SoftBank's chip venture sets sights on AI supremacy.
Key Points:
SoftBank eyes a $100 billion venture into AI chips.
A 3.2% stock uptick follows the venture news.
The project challenges Nvidia, aiming to redefine the semiconductor arena.
Details:
In a similar move to that of OpenAIās Sam Altman, SoftBank has launched Project Izanagi under Masayoshi Son's direction. Son is the founder, CEO and largest shareholder of SoftBank. This $100 billion venture into AI semiconductor technology marks a bold step towards leading the industry potentially trying to counter Sam Altmanās plans. With plans to fund $30 billion of the project itself and seek the remainder from partners in the Middle East, SoftBank is positioning itself as a key player in the next generation of AI advancements. Izanagi, named after a creation deity, symbolizes the ambition to harness technology for a brighter future. This initiative has propelled SoftBank's stock upwards, reflecting confidence in the company's direction after recent successes.
Why It Matters:
Masayoshi Son's latest move give us a clear insight into the challenges and vision of the current tech landscape, promising a shift towards more advanced AI capabilities. It's a testament to SoftBank's strategy of looking beyond the horizon, investing in a future where technology enhances every aspect of our lives.
MAGIC AI
š°ļø Magic AI Secures $117 Million to Build an AI Software Engineer
Evolving AI: Magic AI gets $117M for a new way of coding.
Key Points:
Secured $117M in new funding, total over $145M.
Creating an AI that helps with coding like a team member.
Introducing a breakthrough technology for understanding big projects.
Aiming for AI that works for the good of all.
Details:
Magic AI, from San Francisco, has excited many by raising $117 million. This effort supported by big names like NFDG Ventures, CapitalG, and Elad Gil, is not just about money. It's about a dream started in 2022 to make coding easier and smarter with AI. Imagine an AI that can look at a huge amount of code at once ā much more than what's common today. This is what Magic AI's technology, called LTM Net, does. It's a big leap toward making AI understand and work on large projects, offering a helping hand to developers in a way never seen before.
Our Thoughts:
Magic AI's journey is a bright spot in the tech world. They're not just making coding faster but they're aiming to create AI that understands our big challenges. This could change how we solve problems and make new things. How will this reshape our work and the way we think about technology's possibilities?
š” Tip of the Day
Last week, OpenAI announced Sora, their groundbreaking text-to-video model. OpenAI has compiled a 10-minute compilation of some truly mesmerizing video examples showcasing what this model is capable of. Click below and be ready to be amazed.
šÆ SNAPSHOTS
Direct links to relevant AI articles.
š ElevenLabs: AI sounds effects coming to ElevenLabs, and they showcased it in a clever way using OpenAIās Sora videos to generate them.
šÆ Amazon: Amazon AGI team say their AI is showing āemergent abilitiesā.
š IBM: AI will create more jobs than it destroys.
š OpenAI: OpenAI's quest to trademark "GPT" hits a wall at the US Patent and Trademark Office.
š Trending AI Tools
š¼ Meetingly - Transform sales calls into actionable insights effortlessly (link)
š Lume- Automate data mappings with AI (link)
š„ļø Modelize AI - AI workflows for tailored and fast results (link)
š§® Equals - An AI assistant for your spreadsheet (link)
š Byrdhouse AI 2.0- Multilingual video call AI interpreter (link)
What'd you think of today's edition? |
Reply