AI Rundown by Lightscape Partners – 01/08/24

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

As we find our stride in the new year, let’s look at what happened this week in artificial intelligence:

This week's highlights include OpenAI’s announcement to launch GPT Store for AI apps, a new ChatGPT update hinting at the takeover of Siri and Google Assistant, and Microsoft announcing their first keyboard change in decades to welcome a seamless AI integration.

Top AI Developments of the Week

OpenAI to Launch GPT Store for AI Apps. Link.

OpenAI is set to launch its GPT Store, a platform for custom apps based on its AI models like GPT-4, in the coming week. This initiative will enable developers to create and list GPTs publicly, provided they comply with OpenAI's updated usage policies and brand guidelines.

• To list GPTs in the store, developers must review OpenAI's policies, verify their profiles, and publish GPTs as "public."

• GPTs can range from simple to complex applications, like a recipe assistant or a tool for code style checking.

• OpenAI’s tool allows developers to create AI chatbots by simply typing in desired capabilities.

• As of now, OpenAI has not disclosed a revenue-sharing plan for the GPT Store.

The launch of the GPT Store marks OpenAI's transition from an AI model provider to a platform, democratizing the creation of generative AI apps. This move could significantly impact consultancies that build similar AI solutions and change the landscape of AI app development.

ChatGPT's Update May Challenge Siri and Google Assistant. Link.

ChatGPT's latest update, reported by Android Authority, has introduced a feature that could position it as a competitor to existing digital assistants like Siri and Google Assistant. This development indicates a significant shift in the capabilities of AI-powered digital assistants.

• ChatGPT can now function as a phone's voice assistant, a feature that is disabled by default but can be activated manually.

• The update suggests that ChatGPT might replace traditional digital assistants.

• The feature is not fully operational yet, with limitations outside of the ChatGPT app.

• The code, including com.openai.voice.assistant.AssistantActivity, indicates readiness for ChatGPT to become a default digital assistant.

• OpenAI CEO Sam Altman has spoken about AI changing interactions with technology, and there are AI hardware products in development.

The introduction of this feature in ChatGPT could revolutionize the digital assistant landscape, offering more advanced AI capabilities in everyday devices. It highlights the rapid advancements in AI technology and its potential integration into consumer electronics.

Microsoft Introduces First Change to Keyboard in Decades. Link.

Microsoft has announced the addition of a new Copilot key for AI-powered Windows PCs, marking the first major update to the Windows PC keyboard in nearly 30 years. This key is part of Microsoft's effort to integrate AI seamlessly into Windows, from the system to the hardware.

• The Copilot key, similar to the Windows key, will become a standard component of the PC keyboard.

• AI companion Copilot is baked into the operating system, offering a more personal and intelligent computing experience.

• Copilot was previously announced to be available across all Microsoft 365 applications.

• The Copilot key will start appearing on Windows 11 PCs from late February through Spring, including on new Surface devices.

• Microsoft hinted at further enhancements to Windows to optimize AI experiences, blending local and cloud processing.

This development signifies a major shift in personal computing, with Microsoft leading the way in integrating AI capabilities directly into the operating system and hardware. The Copilot key on Windows 11 could redefine user interactions and productivity with AI-driven functionalities.

Venture

Perplexity AI Raises $73.6 Million to Revolutionize Search. Link.

California-based Perplexity AI, a startup focused on reinventing knowledge discovery, announced a $73.6 million Series B funding round. The company aims to challenge Google and Microsoft in the search engine market with its AI-native platform.

• Perplexity AI's total raise is now $100 million, with an estimated valuation over $500 million.

• The round includes IVP, NEA, Databricks, Nvidia, Bessemer Venture Partners, and Kindred Ventures, along with tech leaders like Jeff Bezos and Shopify CEO Tobi Lutke.

• Perplexity AI provides a conversational search experience, integrating web indexes with AI models, and offers transparent access to information sources.

• The platform is already popular, with 10 million monthly active users and over half a billion queries in 2023.

• A $20/month Pro version offers enhanced model choices, including GPT-4, and additional features like Perplexity Copilot.

Perplexity AI's funding signals a significant move in the search engine industry, focusing on a more efficient, AI-driven approach to information discovery. With its unique model and growing user base, Perplexity AI could redefine how users interact with search engines and access information.

Intel Launches Articul8 AI for Enterprise Software. Link.

Intel is spinning out a new company, Articul8 AI, focusing on generative AI enterprise software. This initiative aligns with Intel's strategy to expand in the AI-powered enterprise software market, offering solutions optimized for Intel hardware.

• Articul8 AI is created in collaboration with DigitalBridge and builds on a proof-of-concept developed with Boston Consulting Group.

• The GenAI system can read text and images, addressing security needs within enterprise environments.

• Articul8 AI's platform caters to sectors like financial services, aerospace, and telecommunications, emphasizing security and specialized domain knowledge.

• Arun Subramaniyan, former VP at Intel, will be the CEO of Articul8 AI, comprising ex-Intel employees.

• The spinout is part of Intel’s broader strategy under CEO Pat Gelsinger, focusing on new software products and competing with firms like Nvidia and AMD.

Articul8 AI represents Intel's significant move into the GenAI software market, potentially enhancing the appeal of Intel hardware through advanced AI capabilities. It marks a strategic shift towards integrating AI software solutions with hardware manufacturing.

LG NOVA and West Virginia's $700M Tech Investment. Link.

LG NOVA, the innovation arm of LG Electronics, has partnered with West Virginia to invest $700 million in tech startups and projects, aiming to drive economic growth and technology advancement in the state.

• LG Electronics and West Virginia will focus on renewable energy, healthcare, and emerging industries.

• The initiative is expected to create 275 new jobs, particularly in tech sectors.

• The NOVA West Virginia Investment Fund plans to inject $700 million over five years into various projects within the state.

• Beyond local growth, LG aims to develop transformative technologies that positively impact the planet.

This collaboration marks a significant economic development for West Virginia, potentially transforming it into a hub for technology and innovation. It reflects a strategic shift towards diversifying the state's economy and fostering a new era of technological advancements.

Hardware + Software

Google DeepMind's New Training Methods for Robots. Link.

Google's DeepMind Robotics team has announced new training methods for robots, AutoRT and RT-Trajectory, to enhance robot understanding and task execution. This research aims to blend generative AI and large language models (LLMs) with robotics for more versatile applications.

• Uses Visual Language Models (VLMs) for situational awareness and can manage a fleet of robots. It suggests tasks aligning with the hardware's capabilities and has managed up to 20 robots and 52 devices in trials.

• This method leverages video input, overlaying a 2D sketch of a robot arm over a video to provide visual hints for robot-control policies. It has shown a double success rate in training compared to previous methods.

• These methods aim to move beyond single-purpose robots, enabling them to adapt to changes and perform a variety of tasks.

• DeepMind's trials have included over 77,000 trials and 6,000 tasks, with RT-Trajectory achieving a 63% success rate in 41 tasks.

DeepMind's research represents a significant leap in robotics, potentially leading to robots that can understand natural language commands and perform diverse tasks with efficiency and accuracy. It opens possibilities for applying AI in novel situations and broadens the scope of robotics in various fields.

Microsoft Copilot Now Available on Mobile Platforms. Link.

Microsoft has released its Copilot app on Android, iOS, and iPadOS, providing users with access to advanced AI capabilities for various creative and professional tasks.

• Copilot, powered by GPT-4 and DALL·E 3, offers functionalities like drafting emails, composing stories, summarizing texts, creating travel itineraries, updating resumes, and more.

• The app includes an Image Creator feature, allowing users to generate designs, logos, backgrounds, and storyboards.

• Since its launch, Copilot has been downloaded over 1.5 million times worldwide on both Android and iOS platforms.

• Copilot is free to use, contrasting with OpenAI's GPT app, which charges for GPT-4 access.

• Formerly known as Bing Chat, Copilot was rebranded in November and was previously available as a web service and through the Bing app.

The release of Copilot on mobile platforms marks a significant expansion of Microsoft's AI services, offering users seamless access to powerful AI tools for creative and professional tasks. This move strengthens Microsoft's position in the AI app market and enhances the accessibility of AI technology for a broader user base.

Regulation + Privacy

California's New Bills for AI Regulation. Link.

California State Senator Steve Padilla has introduced two bills, Senate Bills 892 and 893, aimed at regulating AI systems, especially those used by state agencies. These bills seek to establish safety, privacy, and non-discrimination standards for AI services and a research hub for AI development.

• Requires the Department of Technology to develop AI standards and prohibits the state from contracting AI services that don't meet these standards.

• Proposes the creation of a California AI research hub to support academic research with compute resources and data.

• These bills follow a previous bill by Senator Scott Wiener focused on transparency and security in AI.

• California, a hub for AI companies, has recently passed several tech-focused laws. The state's initiative reflects a broader trend of state-level AI regulation in the absence of major federal laws.

The introduction of these bills signifies a growing effort to establish a legal framework for AI at the state level. It highlights the need for responsible and ethical use of AI in government and public services and could set a precedent for other states and federal legislation.

OpenAI Adapts to EU Data Privacy Regulations. Link.

In response to EU scrutiny over data privacy, OpenAI is updating its terms and privacy policy to align with the European Union's General Data Protection Regulation (GDPR). This includes designating its Irish entity, OpenAI Ireland Limited, as the data controller for EEA and Swiss residents.

• OpenAI faces investigations in Italy and Poland over ChatGPT's data processing and privacy impacts.

• OpenAI's new terms of use, effective February 15, 2024, shift legal responsibilities to its Dublin-based subsidiary.

• The changes are meant to streamline privacy oversight under GDPR's one-stop-shop mechanism.

• This move is part of OpenAI's broader efforts to mitigate regulatory risks and expand its presence in the EU market.

OpenAI's policy update is a strategic effort to comply with stringent EU data privacy laws and reduce the likelihood of unilateral actions by individual EU member state regulators. It underscores the challenges tech companies face in navigating global data privacy standards and the importance of regional compliance.

US SEC Rules on Apple and Disney's AI Usage Shareholder Votes. Link.

The U.S. Securities and Exchange Commission (SEC) has denied requests by Apple and Disney to exclude shareholder votes on their use of artificial intelligence (AI). These votes were proposed by a labor group concerned with AI's impact on professional work and creative rights.

• The largest American labor union federation, AFL-CIO, filed proposals asking for reports on AI usage and ethical guidelines at Apple and Disney.

• The proposals arise from fears that AI might replace creative and professional workers and misuse their work, as seen in recent labor disputes and lawsuits.

• Both companies argued that the proposals pertained to ordinary business operations and could be excluded from ballots.

• The SEC ruled that these proposals transcend ordinary business matters and do not seek to micromanage the companies.

The SEC's decision signifies a growing recognition of the ethical implications of AI in business. It may lead to more transparency and accountability in how major corporations like Apple and Disney use AI technology, aligning with broader disclosures by other tech companies.

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