👀 Show me the data!

🗞️ Your Weekly Round-Up

1️⃣ Gemma

This week, Google released Gemma, their state-of-the-art open-source AI born out of the Gemini project. 

Here’s the TL;DR: 

• Gemma comes in two model weights: 2B and 7B. 

• Both are pre-trained and can run on a developer laptop or desktop with CPU or GPU - and Google Cloud with GPU and TPU acceleration.

• The Gemma 7B model outperforms other open-source models on various benchmarks, such as MMLU and GSM8K.

• Currently, Gemma only offers text-to-text capabilities.

Read more: Mashable

2️⃣ Reddit to license out user-generated content

Reddit just inked a lucrative licensing deal, granting Google access to their user-generated content.

Valued at approximately $60 million annually, this partnership represents a departure from the previous norm, where AI firms often relied on unlicensed data from the open web, facing legal uncertainties. 

This deal dwarfs OpenAI's reported $5 million annual agreements with news publishers and Apple's multi-year, multimillion-dollar deals with prominent news outlets.

Plus, here’s a thought experiment: If Reddit can fetch $60 million for their UGC, how much could Google, Meta, and all the other tech giants command for their data?

Read more: The Verge

3️⃣ Show me the data!

Last week, OpenAI launched Sora, a ground-breaking text-to-video AI model - that left many in the creative community demanding answers regarding its training data sources. 

It’s speculated that Sora was trained on vast amounts of video data scraped from the internet, potentially including copyrighted works. However, OpenAI's reluctance to disclose specifics about Sora's training data has fueled concerns about the legality and ethics of data acquisition for AI models.

The lack of transparency prompted major backlash from artists and creatives, accusing OpenAI of using stolen work to train AI models. 

• Filmmaker Justine Bateman minced no words, condemning Sora's alleged reliance on stolen content as "repulsive."

• Professionals in film VFX, like Jimmy Lanceworth, expressed apprehension about the future of their industry in the wake of AI advancements.

Read more: Mashable

4️⃣ Air Canada’s rogue bot loses court case

In a precedent-setting case, a Canadian court rejected Air Canada's attempt to deflect responsibility for its AI-powered chatbot's actions.

The incident, dating back to 2022, involved Jake Moffatt, who was seeking information on bereavement fare eligibility via Air Canada's chatbot.

Misled by the bot's response, Moffatt was wrongly informed about retroactive refunds, contrary to the airline's policy.

This ruling carries significant implications for companies integrating AI or machine learning into their customer service platforms. 

The biggest lesson? Make sure to thoroughly test your bots before deploying 🤖 

Read more: Mashable

5️⃣ AI 🤝 Crypto

Vitalik Buterin, Co-founder of Ethereum, says that AI is a potential game-changer in addressing one of Ethereum's major technical challenges: identifying bugs and verifying code. 

In a recent post, Buterin expressed his excitement about AI-assisted formal verification, highlighting the need to mitigate the risk of bugs in Ethereum's codebase.

He had also previously discussed the potential for AI-driven agents to revolutionize various aspects of decentralized systems. 

Read more: The Block

6️⃣ Scale AI signs government contract

LLMs, which can transform military planning and decision-making, also pose unprecedented challenges in terms of safety and reliability. 

To address this, the Pentagon’s Chief Digital and Artificial Intelligence Office (CDAO) signed a contract with Scale AI to develop an LLM testing system. 

This one-year deal aims to help the military use AI safely, offering real-time feedback for warfighters and specialized evaluation sets tailored to military applications.

Read more: Defenses Scoop

7️⃣ Adobe’s AI Assistant

AI assistants are about to become ubiquitous in big tech.

This week, Adobe released their AI Assistant, a GenAI experience designed to streamline document navigation and comprehension. 

Here’s the TL;DR: 

• Users can interact with its conversational engine to search through documents and extract key information.

• It promises to simplify tasks like research, report summarization, and content formatting. 

• Compatible with various document formats, including Word and PowerPoint, the chatbot ensures data security by sticking to Adobe’s protocols.

Read more: The Verge

8️⃣ AI safety and alignment

DeepMind, the AI R&D arm behind Google's GenAI projects, recently formed the AI Safety and Alignment organization to focus on ensuring the responsible development and deployment of AI technologies - particularly their Gemini and Gemma models. 

Led by Anca Dragan, a seasoned expert in AI safety, the organization will tackle issues like: 

• Preventing misinformation, 

• Safeguarding against biased outputs, and

• Addressing the broader implications of AI in society.

Read more: TechCrunch

🔍️ SciSpace - The #1 search engine curated for Science

How do you conduct academic research efficiently in a world with long literature reviews and complex jargon, among other scientific hindrances?

You use SciSpace. 

With SciSpace, you can find 282 million+ relevant papers for any topic or question and analyze and contrast multiple papers on one screen.

Furthermore, you can ask Copilot for summaries, explanations, or questions to read papers faster.

You can also save all your notes and get writing suggestions with the Notebook feature. 

Get unlimited access to all the best features of SciSpace:

• Use coupon code AUTO40 for 40% off.

• Use AUTO20 to get 20% off on a monthly plan.

🎙️ New Episode ft. Otto von Zastrow-Marcks, Founder and CEO of Midpage

Hey team! This week, we had the pleasure of having Otto von Zastrow-Marcks, Founder and CEO of Midpage, on the pod.

During the episode, he discussed:

• Why auto-completion is difficult to carry out in legal tech,

• How Midpage turns snippets into arguments,

• Building the right UI/UX,

• The biggest challenge of working with non-deterministic code,

• The future of legal tech,

• And much more!

Catch the episode on YouTube, Spotify, or Apple Podcasts today!

🤑 AI Fundraising News

• Athos raised $35M in Series B funding to support its AI-based precision small molecules for autoimmune diseases and various cancer indications.

• Rasa raises $30M in Series C funding to accelerate its drive to enable businesses to use generative conversational AI.

• Magic AI raises $117M in Series B funding to build an AI software engineer to assist with complex coding tasks.

• Altavo raises $5.4M/€5M in Series A funding to develop a medical device called Artificial Voice that uses proprietary "silent speech" technology to help voiceless individuals regain a natural-sounding voice.

• Recogni raises $102M in Series C funding to develop AI inference solutions that improve performance and efficiency while reducing costs. The company specializes in AI-based inference processing solutions for Generative AI and autonomous platforms, and has already introduced its first low-power AI compute product, Recogni Scorpio.

• BRIA raises $24M in Series A funding to provide a responsible visual generative AI open platform that allows developers to integrate generative AI capabilities into existing products, solutions, or systems.

• Moonshot AI raises $1B in Series B funding for an LLM focused on long-form context and response, an area that has long outsmarted others in the field.

• Napier AI raises $57M/£45M in funding to develop AI solutions that help customers identify criminal or suspicious activities and provide regulatory reporting, offering trusted technology and delivery to financial institutions.

• Simetrik raises USD 55M in Series B funding to expand its global reach and develop solutions to automate key financial tasks using its scalable Simetrik Building Blocks (SBBs) based on generative AI technologies.
  
  
  

🛠️ AI Tools to Check Out

• Ioni: A smart chatbot based on the latest AI technology that talks like a human and creates CTA for your customers. Check them out!

• Uizard: An AI-powered UI design tool built to empower product teams to ideate, design, and iterate faster than ever before. Check them out!

🐦 Tweet Post of the Day

📰 Lenny’s weekly newsletter had some fascinating examples of how companies use custom GPTs. Check them out here!