Qwen 1.5: Alibaba's 72 B, Multilingual Gen AI Model
Plus: AI software reading ancient scrolls, Robolox's new chat translation system
Hello Engineering Leaders and AI Enthusiasts!
Welcome to the 204th edition of The AI Edge newsletter. This edition brings you Alibaba's multilingual Gen AI model Qwen 1.5 and how it is competing with GPT-3.5.
And a huge shoutout to our amazing readers. We appreciate you😊
In today’s edition:
🆕Qwen 1.5: Alibaba's 72 B, multilingual Gen AI model
🏛️ AI software reads ancient words unseen since Caesar's era
⌚️ Roblox users can chat cross-lingually in milliseconds
📚 Knowledge Nugget: How to use generative AI for historical research by
Let’s go!
Qwen 1.5: Alibaba's 72 B, multilingual Gen AI model
Alibaba has released Qwen 1.5, the latest iteration of its open-source generative AI model series. Key upgrades include expanded model sizes up to 72 billion parameters, integration with HuggingFace Transformers for easier use, and multilingual capabilities covering 12 languages.
Comprehensive benchmarks demonstrate significant performance gains over the previous Qwen version across metrics like reasoning, human preference alignment, and long-context understanding. They compared Qwen1.5-72B-Chat with GPT-3.5, and the results are shown below:
The unified release aims to provide researchers and developers an advanced foundation model for possible downstream applications. Quantized versions allow low-resource deployment. Overall, Qwen 1.5 represents steady progress towards Alibaba's goal of creating a "truly 'good'' generative model aligned with ethical objectives.
Why does this matter?
This release signals Alibaba's intent to compete with Big Tech firms in steering the AI race. The upgraded model enables researchers and developers to create more capable assistants and tools. Qwen 1.5's advancements could enhance education, healthcare, and sustainability solutions.
AI software reads ancient words unseen since Caesar's era
Nat Friedman (former CEO of Github) uses AI to decode ancient Herculaneum scrolls charred in the 79AD eruption of Mount Vesuvius. These unreadable scrolls are believed to contain a vast trove of texts that could reshape our view of figures like Caesar and Jesus Christ. Past failed attempts to unwrap them physically led Brent Seales to pioneer 3D scanning methods. However, the initial software struggled with the complexity.
A $1 million AI contest was launched ten months ago, attracting coders worldwide. Contestants developed new techniques, exposing ink patterns invisible to the human eye. The winning method by Luke Farritor and the team successfully reconstructed over a dozen readable columns of Greek text from one scroll. While not yet revelatory, this breakthrough after centuries has scholars hopeful more scrolls can now be unveiled using similar AI techniques, potentially surfacing lost ancient works.
Why does this matter?
The ability to reconstruct lost ancient knowledge illustrates AI's immense potential to reveal invisible insights. Just like how technology helps discover hidden oil resources, AI could unearth 'info treasures' expanding our history, science, and literary canons. These breakthroughs capture the public imagination and signal a new data-uncovering AI industry.
Roblox users can chat cross-lingually in milliseconds
Roblox has developed a real-time multilingual chat translation system, allowing users speaking different languages to communicate seamlessly while gaming. It required building a high-speed unified model covering 16 languages rather than separate models. Comprehensive benchmarks show the model outperforms commercial APIs in translating Roblox slang and linguistic nuances.
The sub-100 millisecond translation latency enables genuine cross-lingual conversations. Roblox aims to eventually support all linguistic communities on its platform as translation capabilities expand. Long-term goals include exploring automatic voice chat translation to better convey tone and emotion. Overall, the specialized AI showcases Roblox’s commitment to connecting diverse users globally by removing language barriers.
Why does this matter?
It showcases AI furthering connection and community-building online, much like transport innovations expanding in-person interactions. Allowing seamless cross-cultural communication at scale illustrates tech removing barriers to global understanding. Platforms facilitating positive societal impacts can inspire user loyalty amid competitive dynamics.
Enjoying the daily updates?
Refer your pals to subscribe to our daily newsletter and get exclusive access to 400+ game-changing AI tools.
When you use the referral link above or the “Share” button on any post, you'll get the credit for any new subscribers. All you need to do is send the link via text or email or share it on social media with friends.
Knowledge Nugget: How to use generative AI for historical research
Can AI be a historian's research assistant instead of a replacement? In his detailed and interesting piece Historian
explores how AI could augment primary source research rather than replace human skills. Built on GPT-4, "Historian's Friend" had been tested on materials from different eras. Benjamin validates its usefulness in accelerating processes like translation, visualization and search.The Historian’s Friend helped in visualizing and quantifying data from 1930s magazines, translating text from obscure medical texts. While it couldn't crack CIA redactions yet, its web integration enabled quick biographies. However, it still made errors that required human expertise to catch. Breen makes the case that by automating rote processes, AI could enable more people to undertake historical research, while historians focus on providing wisdom, oversight and moral compass.
Why does this matter?
With AI, we can augment human abilities rather than replace them. The experiments with historical research shows how AI tools can enhance productivity and insight - from faster translation and transcription, to surfacing unseen connections in images and data. When combined with uniquely human skills, it can provide value and advantage.
What Else Is Happening❗
📰 Semafor tests AI for responsible reporting
News startup Semafor launched a product called Signals - AI-aided curation of top stories by its reporters. An internal search tool helps uncover diverse sources in multiple languages. This showcases responsibly leveraging AI to enhance human judgment as publishers adapt to changes in consumer web habits. (Link)
🕵️♂️ Bumble's new AI feature sniffs out fakes for safer matchmaking
Bumble has launched a new AI tool called Deception Detector to proactively identify and block fake profiles and scams. Testing showed it automatically blocked 95% of spam accounts, reducing user reports by 45%. This builds on Bumble's efforts to use AI to make its dating and friend-finding platforms safer. (Link)
⚙️ Huawei repurposes factory to prioritize AI chip production over its bestselling phones
Huawei is slowing production of its popular Mate 60 phones to ramp up manufacturing of its Ascend AI chips instead, due to growing domestic demand. This positions Huawei to boost China's AI industry, given US export controls limiting availability of chips like Nvidia's. It shows the strategic priority of AI for Huawei and China overall. (Link)
💷 UK to spend $125M+ to tackle challenges around AI
The UK government will invest over $125 million to support responsible AI development and position the UK as an AI leader. This will fund new university research hubs across the UK, a partnership with the US on the responsible use of AI, regulators overseeing AI, and 21 projects to develop ML technologies to drive productivity. (Link)
🤝 Europ Assistance partnered with TCS to boost IT operations with AI
Europ Assistance, a leading global assistance and travel insurance company, has selected TCS as its strategic partner to transform its IT operations using AI. By providing real-time insights into Europ Assistance's technology stack, TCS will support their business growth, improve customer service delivery, and enable the company to achieve its mission of providing "Anytime, Anywhere" services across 200+ countries. (Link)
New to the newsletter?
The AI Edge keeps engineering leaders & AI enthusiasts like you on the cutting edge of AI. From machine learning to ChatGPT to generative AI and large language models, we break down the latest AI developments and how you can apply them in your work.
Thanks for reading, and see you tomorrow. 😊