Bet You've Never Heard of Deep Search

AI's Carbon Footprint

A recent study reported by MIT Technology Review reveals the significant energy consumption and carbon emissions linked to using generative AI models. Generating a single image with these advanced AI models is as energy-intensive as fully charging a smartphone. While generating text is less energy-consuming than creating images, the study underscores that the majority of the carbon footprint of these massive AI models comes from their use, not just their training. This finding emphasizes the urgent need to be selective in deploying generative AI and to consider more specialized, less energy-intensive alternatives. It also calls for greater awareness of the environmental impact of AI usage and urges both consumers and companies to be accountable for their energy consumption and emissions.

Why it matters: This study serves as a crucial reminder of the environmental costs associated with AI technology and the importance of mindful usage to minimize carbon emissions. Understanding the energy and carbon implications of generative AI is vital for balancing technological advancement with environmental responsibility, ensuring a sustainable future in the rapidly evolving AI landscape.

Source

AstraZeneca Partners with AI Firm Absci for Cancer Drug Development

AstraZeneca, known for its COVID-19 vaccine production, has forged a $247 million collaboration with American AI biologics firm Absci to harness the power of generative AI in designing a cancer-fighting antibody. Absci's innovative zero-shot generative AI model focuses on optimizing drug candidates based on factors such as target affinity, safety, and manufacturability.

Why it matters: This collaboration holds great promise in advancing cancer therapeutics, leveraging AI's capabilities to expedite drug discovery.

Source

AI Alliance of Over 50 Companies, Including IBM and Meta, Aims to Disrupt AI Dominance

IBM, Meta, and a consortium of more than 50 companies have come together to establish the AI Alliance, a coalition aimed at challenging the dominance of OpenAI, Microsoft, Google, and Amazon in the field of artificial intelligence. This diverse alliance, encompassing chip manufacturers, AI startups, and prestigious universities, seeks to champion open-source AI development, foster common frameworks for AI algorithm evaluation, and allocate resources to AI research. By doing so, they aim to counter the perception of tech giants' dominance and encourage a more collaborative and transparent approach to AI innovation, ultimately benefiting the entire field.

Why it matters: The formation of the AI Alliance signifies a noteworthy shift in the AI landscape, with major industry players uniting to disrupt the status quo and advocate for openness and collaboration in AI development. This collective effort promises to challenge the stronghold of OpenAI and tech giants, potentially catalyzing a wave of innovation and progress within the AI community.

Source

Microsoft Bing's 'Deep Search' with GPT-4 Elevates Search Precision and Comprehensiveness

Microsoft Bing has introduced a groundbreaking feature known as "Deep Search," driven by OpenAI's GPT-4, aimed at delivering more thorough responses to intricate search queries. Deep Search augments Bing's existing web index and ranking system by expanding the search query into a comprehensive description of desired results, enabling users to articulate their intent more precisely. When faced with ambiguous queries, Deep Search identifies all potential intents, providing users with multiple comprehensive descriptions to choose from. This feature subsequently retrieves pertinent results that may not typically appear in standard search results, optimizing the search process and producing tailored outcomes based on a user's specific needs.

Why it matters: Deep Search marks a significant advancement in the search engine domain, enriching users' search experiences by offering more in-depth and personalized results for complex queries. By leveraging OpenAI's GPT-4, Microsoft Bing enhances its ability to comprehend user intent and generate comprehensive descriptions, promoting accuracy, relevance, and trustworthiness in search results.

Source

Microsoft Unveils Major Enhancements to Copilot with GPT-4 Turbo, DALL-E 3, and More

Microsoft is rolling out substantial updates to its Copilot service, including integration with OpenAI's GPT-4 Turbo model, an improved DALL-E 3 model for image creation, a code interpreter feature, and deep search functionality within Bing. Copilot will soon leverage the power of GPT-4 Turbo, offering a larger 128K context window for more precise understanding and responses to queries. The updated DALL-E 3 model enhances image quality and accuracy in Bing Image Creator and Copilot. Microsoft Edge users can compose text within website inputs and summarize YouTube videos with a Copilot sidebar. The new code interpreter feature facilitates complex natural-language requests for calculations, data analysis, and code generation. Users can also upload and download files to and from Copilot.

Why it matters: Microsoft's continuous improvement and integration of advanced AI models like GPT-4 Turbo and DALL-E 3 into Copilot represent a significant leap in the capabilities of AI-assisted tools for users across various domains, from content creation to coding and data analysis.

Source

Google Unveils Game-Changing AI Model 'Gemini,' Set to Challenge OpenAI's Dominance​

In a groundbreaking move (and a complete 180 since the last newsletter release), Google has unveiled its highly anticipated "family of multimodal AI models" known as Gemini. Comprising three distinct models—Gemini Ultra, Gemini Pro, and Gemini Nano—Gemini Ultra, in particular, has left an indelible mark by achieving state-of-the-art results in 30 out of 32 LLM benchmarks, surpassing GPT-4 and other language models in image understanding, video comprehension, and speech recognition. While a release date for Gemini Ultra remains unspecified, Google has introduced Gemini Pro, on par with GPT-3.5, integrated into Bard, their chatbot. Additionally, Gemini Nano, designed for mobile devices, will join the Pixel series next year. This announcement has positioned Google as a formidable contender in the race for AI dominance, challenging OpenAI's leading role heading into 2024.

Why it matters: Google's introduction of the Gemini AI model represents a significant shift in the AI landscape, with the potential to disrupt OpenAI's supremacy. Gemini's multimodal capabilities, spanning text, code, images, audio, and video, herald new possibilities for AI integration into everyday tasks and applications, ushering in an era of enhanced AI-driven solutions.

Source

Meta Launches 28 Celebrity-Inspired AI Characters on WhatsApp, Messenger, and Instagram

Meta, the parent company of social media giants like Facebook, has officially rolled out 28 AI characters on its platforms, including WhatsApp, Messenger, and Instagram. These AI avatars, such as Billie (inspired by Kendall Jenner), Zach (modeled after Mr. Beast), and Bru (based on Tom Brady), are now accessible for conversations at any time. Users can take advantage of the "long-term memory" feature, which allows them to retrieve past conversations with their favorite celebrity-inspired AI avatars. This move raises the question of whether people are open to engaging in daily conversations with AI avatars resembling celebrities.

Why it matters: Meta's introduction of celebrity-inspired AI characters reflects the company's ongoing efforts to enhance user engagement and personalization across its platforms. This development could potentially open up new avenues for virtual interactions and entertainment, blurring the lines between human and AI interactions. As AI avatars become more sophisticated and capable of retaining past conversations, it raises intriguing possibilities for the future of social media and virtual relationships, further intertwining technology and celebrity culture.

Source

Apple Quietly Unveils MLX: A Powerful Machine Learning Framework

Apple has stealthily introduced MLX, a machine learning framework aimed at enabling developers to construct AI models optimized for Apple's proprietary silicon processors. This marks Apple's initial foray into the AI realm. MLX is engineered for user-friendly operation and robust capabilities, allowing for the training of intricate AI models. While it may take another year or two for Apple to fully venture into the AI arena with a consumer-oriented offering, the company's reputation for innovation and quality assures that it will make a notable impact when it does.

Why it matters: Apple's quiet release of MLX underscores the company's growing interest in the AI field, aligning with its broader strategy to leverage in-house-designed processors for enhanced performance and efficiency. This move hints at Apple's potential future AI products or services, promising a seamless integration of AI capabilities into its ecosystem.

Source

Meta Unveils 'Imagine with Meta,' a DALL-E Competitor for AI-Generated Images

Meta, the company behind Facebook and Instagram, has introduced "Imagine with Meta," a standalone image generator powered by Meta's Emu AI model. This tool enables users to create high-resolution images based on text descriptions, similar to the capabilities of DALL-E and MidJourney. Initially available for free in the U.S., it allows for the generation of up to four images per prompt and incorporates an invisible watermark to ensure authenticity, transparency, and traceability. This move aligns with the industry trend of watermarking AI-generated content, helping users distinguish between human and AI-created materials.

Why it matters: Meta's launch of "Imagine with Meta" represents a significant step in the AI-driven image generation landscape, putting the company in competition with existing AI models like DALL-E. This development showcases Meta's dedication to remaining at the forefront of AI technology and providing users with innovative tools for content creation and personalization. The inclusion of watermarking emphasizes the importance of transparency in AI-generated content, ensuring ethical usage and credibility.

Try it out HERE

Source

Prompts to Try - Get Gen_ID 📷

If you create an image in DALL-E 3, and you want to keep that style for future prompting, type in the prompt “Get Gen_ID”. You will get a list of random letters and numbers (e.g. 4yuG4HZyCHwGsXsx).

You can then enter the following prompt, entering the Gen_ID of your choice:

photo of [your subject/object] --style [Gen_ID] --ar 16:9