The AI Rundown by Lightscape Partners – 01/16/24

Good morning and welcome to this week's edition of the AI Rundown, brought to you by Lightscape Partners.

This week’s highlights include the release of OpenAI’s GPT Store, OpenAI’s announcement on combatting misinformation in upcoming elections, and Microsoft’s expansion of its copilot.

Join us as we explore these pivotal developments that are shaping the future of artificial intelligence across various sectors.

Top AI Developments of the Week

OpenAI Launches GPT Store for Custom Chatbots. Link.

OpenAI has officially opened its GPT Store, a platform where developers can share custom chatbots they've created. This development signifies a major step in expanding the use and applications of ChatGPT beyond OpenAI's initial offerings.

• The store features chatbots across categories like writing, research, programming, and lifestyle, including notable ones for hiking trail recommendations, code tutoring, and design.

• One of the chatbots, Consensus, offers access to over 200 million academic papers.

• While developers cannot directly charge for their chatbots, the GPTs are available to all paid ChatGPT users. OpenAI plans to share subscription revenue with developers of popular GPTs, though specific details are yet to be provided.

• OpenAI also introduced ChatGPT Team, a new subscription plan targeted at small businesses, offering features like a shared workspace and collaboration on custom GPTs.

The GPT Store's launch is akin to the iPhone's App Store in its potential to foster a wide range of applications and uses for AI technology. It could significantly broaden the scope of ChatGPT's utility and spur innovation in AI-driven applications.

OpenAI's Strategy to Tackle Election Misinformation. Link.

OpenAI has updated its policies to address potential election misinformation during the 2024 elections, acknowledging the challenges posed by AI tools like ChatGPT and Dall-E in spreading false information.

• OpenAI now prohibits the use of its tools for impersonating candidates or local governments, engaging in campaigns or lobbying, discouraging voting, or misrepresenting the voting process.

• Plans to incorporate the Coalition for Content Provenance and Authenticity’s (C2PA) digital credentials into images generated by Dall-E to trace their origins.

• OpenAI's tools will redirect U.S. voting questions to CanIVote.org for authoritative information on voting processes.

• The effectiveness of these measures partly relies on users reporting misuse of AI tools.

OpenAI's proactive steps are crucial for safeguarding democratic processes against AI-generated misinformation. However, the success of these efforts will depend on both technological solutions and increased public awareness and reporting.

Microsoft Expands Copilot AI Access to All Businesses. Link.

Microsoft has made its AI-powered Office assistant, Copilot for Microsoft 365, generally available to all businesses, removing the previous minimum user requirement and opening up access for small and medium-sized businesses.

Key Updates:

1. User Requirement Removal: The 300-seat purchase requirement has been eliminated, allowing businesses of any size to access Copilot for Microsoft 365.

2. Plan Inclusivity: Copilot is now available not only to Microsoft 365 plans but also to Office 365 E3 and E5 customers. Microsoft 365 Business Standard or Business Premium subscribers can also purchase it.

3. Pricing: Copilot for Microsoft 365 is available at $30 per user, per month.

Copilot Features in Office Apps:

- The AI assistant appears as a sidebar chatbot in Microsoft 365 apps.

- Offers features like text generation in Word, PowerPoint presentation creation from Word documents, and assistance with Excel's PivotTables.

- In Teams, it can summarize meetings, and in Outlook, it can summarize email threads and draft responses.

Different Copilot Versions:

Regular Copilot: Free chatbot available for consumers and businesses.

Copilot Pro: New consumer option at $20 per month, offering AI-powered features in Office apps.

Copilot for Microsoft 365: Premium business subscription with more features at $30 per user

Microsoft Expands Copilot AI Access to All Businesses. Link.

Microsoft has made its AI-powered Office assistant, Copilot for Microsoft 365, generally available to all businesses, removing the previous minimum user requirement and opening up access for small and medium-sized businesses.

• The 300-seat purchase requirement has been eliminated, allowing businesses of any size to access Copilot for Microsoft 365.

• Copilot is now available not only to Microsoft 365 plans but also to Office 365 E3 and E5 customers. Microsoft 365 Business Standard or Business Premium subscribers can also purchase it.

• Copilot for Microsoft 365 is available at $30 per user, per month.

Copilot Features in Office Apps:

• The AI assistant appears as a sidebar chatbot in Microsoft 365 apps.

• Offers features like text generation in Word, PowerPoint presentation creation from Word documents, and assistance with Excel's PivotTables.

• In Teams, it can summarize meetings, and in Outlook, it can summarize email threads and draft responses.

Different Copilot Versions:

• Regular Copilot: Free chatbot available for consumers and businesses.

• Copilot Pro: New consumer option at $20 per month, offering AI-powered features in Office apps.

• Copilot for Microsoft 365: Premium business subscription with more features at $30 per user.

This expansion allows businesses of all sizes to leverage AI capabilities in their daily operations, potentially improving productivity and efficiency.

Venture

Robotics Startup 1X Receives $100M in Funding and Backing by OpenAI. Link.

Norwegian robotics startup 1X Technologies, backed by OpenAI, has raised $100 million in Series B funding. The round included significant investors like Samsung and EQT Ventures, positioning the company to advance its bipedal android, NEO, and support its enterprise customers.

• The $100 million funding comes from EQT Ventures, Samsung NEXT, Nistad group, Sandwater, and Skagerak Capital.

• 1X plans to use the funds to bring its new bipedal android NEO to the market and support existing enterprise customers.

• Originally Halodi Robotics, 1X has developed high-torque servo motors and partnered with OpenAI to integrate AI models for enhanced robot intelligence.

• Its wheeled robot, Eve, is already deployed in industrial settings, and the humanoid NEO is designed for domestic assistance.

This significant funding round highlights the increasing interest and investment in AI-driven robotics. 1X's progress in developing humanoid robots capable of natural language understanding and task execution showcases the potential for robots to work alongside humans in various settings.

Quora Secures $75M Funding for AI Chat Platform Poe. Link.

Quora, the question-and-answer website, has raised $75 million from Andreessen Horowitz (a16z) to expand Poe, its AI chatbot platform. This marks the company's first funding round in almost seven years.

• Quora aims to develop a unique subset of the creator economy centered around AI chatbots, allowing creators to earn revenue from bot and prompt creation.

• Poe offers access to a variety of AI models, including ChatGPT, DALL-E 3, Claude 2, and Stable Diffusion, providing a broad toolkit for creators.

• espite the new funding, Quora's valuation has decreased to $500 million, reflecting recent market shifts and higher capital costs.

• Poe has shown significant growth since its launch, with over 18.4 million installs and 1.22 million monthly active users by October. Quora itself reported its highest usage week recently and boasts over 400 million monthly unique visitors.

Regulation + Legislation

OpenAI Responds to New York Times Copyright Lawsuit. Link.

OpenAI has countered the New York Times' copyright lawsuit against them, asserting that the accusations of ChatGPT misusing the newspaper's content are unfounded. OpenAI claims the Times "intentionally manipulated" ChatGPT to support their lawsuit.

• The New York Times sued OpenAI on December 27, alleging that the company used its articles without permission to train ChatGPT, seeking billions in damages.

• In a blog post, OpenAI argued that the Times' portrayal is not accurate, suggesting the newspaper may have manipulated ChatGPT to produce specific responses.

• This lawsuit is significant in the copyright law sphere, as it could set precedents for AI technology's use of copyrighted material.

The lawsuit and OpenAI's response highlight the legal complexities surrounding AI and copyright, with potential ramifications for how AI companies can use existing content for model training. The outcome of this case could influence future regulations and practices in the AI industry.

EU Scrutinizes Microsoft's Investments in OpenAI. Link.

The European Commission is investigating Microsoft's investments in OpenAI to determine if they conflict with the EU's competition rules, particularly in light of recent turmoil within OpenAI and the close ties between the two companies.

• Microsoft has invested heavily in OpenAI, initially with $1 billion in 2019 and an additional $10 billion in early 2023.

• The European regulator's interest has been piqued following significant internal changes at OpenAI, including the temporary removal of CEO Sam Altman.

• The investigation aims to assess the impact of partnerships between large digital market players and AI developers on market dynamics.

• The Commission is considering whether Microsoft's investment might be subject to review under the EU Merger Regulation.

• Similar investigations are being conducted by US and UK regulators.

This investigation by the European Commission into Microsoft's investments in OpenAI reflects growing global regulatory attention on big tech investments in AI. The outcome could influence future partnerships and investments in the AI sector, particularly in terms of competition and market dynamics.

Software + Hardware

EleutherAI To Make Enhanced Pile Dataset for AI Training ‘Significantly Better’. Link.

EleutherAI, a nonprofit research group, is expanding its large text corpora, the Pile, to create a more extensive and refined dataset for training large language models (LLMs). This initiative comes amid legal and ethical concerns about the data used in popular LLMs.

• The updated Pile dataset will be larger and "substantially better" than the previous version, including new and more diverse data.

• With experience from training several LLMs, EleutherAI plans better data cleaning to make the dataset more suitable for LLM training.

• The new dataset will contain more books and diverse non-academic non-fiction, aiming for broader representation.

• EleutherAI faces a lawsuit for allegedly using copyrighted books in its dataset, highlighting the complex legal landscape of AI training data.

• The organization is addressing issues related to copyright and data licensing, ensuring the use of public domain data, Creative Commons licensed text, and openly licensed scientific articles.

The expansion and refinement of the Pile dataset are pivotal in advancing AI research, offering a more robust foundation for LLM training. It also reflects the ongoing efforts to balance innovation with legal and ethical considerations in AI development.

MMGuardian Launches AI-Powered Smartphone for Kids. Link.

MMGuardian, a company focused on child safety technology, has introduced an AI-powered smartphone for kids in collaboration with Samsung. The announcement was made at CES 2024, highlighting the device's enhanced parental control features and safety measures.

• The smartphone uses AI to detect inappropriate content, such as selfies, to protect against online risks like sexting and cyberbullying.

• The device alerts both parents and children to potential online threats.

• Founded in 2012, MMGuardian has been recognized for its contributions to child safety technology, including features for scanning pictures, texts, and chat messages.

• The partnership with Samsung, established in 2023, led to the beta launch of MMGuardian phones.

• MMGuardian has issued over 10 million alerts to parents since its inception.

The launch of MMGuardian's AI smartphone marks a significant development in child safety technology, offering parents advanced tools to monitor and protect their children's digital interactions. This initiative represents a key advancement in integrating child safety features into consumer electronics.

Google Cloud Introduces GenAI Products for Retail. Link.

At the National Retail Federation’s annual conference in NYC, Google Cloud announced new generative AI (GenAI) products aimed at enhancing online shopping experiences and optimizing back-office operations for retailers.

Key Products:

1. Conversational Commerce Solution: This tool allows retailers to implement GenAI-powered agents on their websites and apps. These agents can converse with customers, offering personalized product suggestions.

2. Catalog and Content Enrichment Toolset: Utilizes GenAI models to generate product descriptions, metadata, and categorization suggestions from minimal inputs. It also enables retailers to create new product images or use descriptions to generate AI-rendered product visuals.

• Concerns have been raised about the accuracy and reliability of AI-generated product descriptions and images.

• Google Cloud’s managing director of retail, Amy Eschliman, emphasized ongoing tool improvements and the importance of human review in the workflow to ensure quality and mitigate risks.

• There's a growing interest in GenAI within the retail sector, with large retailers like Walmart and Amazon already investing in these technologies.

• Google Cloud's poll indicates a significant portion of retail decision-makers feel an urgency to adopt GenAI.

Despite potential benefits, the retail industry's widespread adoption of GenAI solutions remains uncertain due to concerns about quality and effectiveness in previous implementations.

OpenAI Introduces ChatGPT Team for Small and Medium Businesses. Link.

OpenAI has launched 'ChatGPT Team', a new subscription plan for its AI-powered chatbot, ChatGPT, specifically designed for small to medium-sized teams.

Key Features:

1. Team Workspace: The plan offers a dedicated workspace for teams of up to 149 members.

2. Access to Latest Models: Subscribers get access to GPT-4, GPT-4 with Vision, and DALL-E 3, along with file analysis and editing tools.

3. Custom GPT Creation: Teams can build and share custom GPTs for various applications, such as analyzing codebases or generating content.

4. Privacy Assurance: OpenAI ensures not to train models on team data or conversations.

5. Pricing: ChatGPT Team is priced at $30 per user per month or $25 per user per month billed annually.

• ChatGPT Team is more affordable than ChatGPT Enterprise, which costs up to $60 per user per month for a minimum of 150 users.

• It offers more features compared to the individual premium plan, ChatGPT Plus, priced at $20 per month.

The launch caters to a significant market demand, as a survey from ResumeBuilder indicates that 49% of companies use ChatGPT for tasks like coding and content creation, and 30% plan to use it in the future.

ChatGPT Team aims to provide cost-effective, team-oriented AI tools to small and medium businesses, potentially enhancing productivity and operational efficiency in these enterprises.

CES 2024

AI Products at CES. Link.

At CES 2024, various AI products stood out for their practical and helpful applications, ranging from voice synthesis for those with speech difficulties to advanced fertility tracking and elder care solutions.

1. Whispp: This technology synthesizes voices for people who have difficulty speaking due to conditions like throat cancer. It turns whispers into full speech, using speech recognition models that work with whispers and affected speech.

2. Louise: A French startup focusing on fertility tracking and advice. It uses machine learning to analyze patient data for biomarkers that could improve fertility. Louise also introduced its B2C app, Olly, which guides individuals through their fertility journey.

3. Rabbit r1: An AI assistant designed for people with vision impairments, capable of operating phone or web apps beyond basic assistant tasks. It uses a ChatGPT-like voice interface for broader functionality.

4. ElliQ: Devices for elder care that use large language models for more natural conversation, assisting older individuals in staying engaged and ensuring safety and care in assistive facilities.

Other

OpenAI Negotiating Content Licensing with Major News Outlets. Link.

OpenAI is reportedly in discussions with prominent media organizations like CNN, Fox Corp., and Time, alongside the News/Media Alliance, to license news content. This move aims to enhance the accuracy and relevance of OpenAI's AI products, including ChatGPT.

• OpenAI's potential deals could involve licensing articles, videos, and images from these media outlets.

• OpenAI has reportedly proposed licensing fees ranging from $1 million to $5 million annually to publishers for their articles.

• The effort is part of OpenAI's strategy to use publisher content to train ChatGPT and include it in OpenAI's products.

• OpenAI faces a lawsuit by the New York Times for allegedly misusing its intellectual property.

• Time’s CEO Jessica Sibley expressed optimism about reaching an agreement with OpenAI that acknowledges the value of their content.

These negotiations signify OpenAI's active pursuit of partnerships to legally source content for training its AI models, amidst growing concerns about compensation and intellectual property rights in the AI industry.

Thank you for reading the AI Rundown by Lightscape Partners. Stay tuned for the latest updates in the world of Artificial Intelligence!