Tips

May 9, 2024

5 mins

This Week in AI

Hello, Niuralogists!

In the ever-evolving realm of artificial intelligence, this week's edition is committed to delivering the most recent breakthroughs. Our central focus is to analyze how these advancements impact different facets of our lives, including workplaces, businesses, policies, and personal experiences. In this issue, we will explore some updates such as Apple's Enhanced AI Features for iPad and Meta's Introduction of Image and Text Generation for Advantage+ to streamline ad creation processes.

For a more in-depth understanding, keep on reading…

Apple Introduces Enhanced AI Features for iPad

Apple's recent unveiling of its latest iPad lineup in Cupertino, CA, introduced a significant upgrade centered around AI advancements. Highlighting the event was the debut of the M4 chip, specifically designed to enhance AI functionalities, as well as a range of AI-driven features. The larger iPad Pro now boasts the M4 chip, equipped with an enhanced Neural Engine, hailed by CEO Tim Cook as a formidable force in AI processing. With a capacity of 38T operations per second, quadrupling the performance of its predecessors, the M4 chip facilitates the execution of sophisticated AI tasks. Notable additions include a True Tone Flash for document scanning and an array of video, image, and music editing tools, all powered by AI algorithms. Anticipation mounts as reports hint at Apple's ongoing development of AI chips for data centers, dubbed "Project ACDC.

Meta Unveils Image and Text Generation for Advantage+ to Expedite Ad Creation

Meta has enhanced its Advantage+ offering by introducing complete image and text generation capabilities, empowering advertisers to swiftly create engaging and imaginative ads. This move signifies Meta's ongoing commitment to automating aspects of ad creation, aimed at enhancing resource efficiency and ad performance. These new functionalities, set to launch globally later this year, enable advertisers to generate various image variations and overlay text inspired by their original creative concepts. By leveraging AI, Meta aims to streamline the ad creation process, providing options for font types, aspect ratio adjustments, and even tailoring text to reflect the advertiser's brand voice. Despite concerns regarding the role of designers and marketers in this automated process, Meta clarifies that the AI will augment existing assets rather than replace them entirely. Additionally, Meta is actively addressing concerns surrounding licensing and copyright issues associated with AI-generated content. These image and text generation features, alongside Meta's suite of AI tools, will be accessible within Advantage+ Creative, facilitating a more efficient campaign creation process for advertisers.

Source: Meta

Microsoft Conducts Extensive In-House Large Language Model Training

Microsoft is in the process of developing a substantial in-house language model (LLM) dubbed MAI-1, boasting a massive 500 billion parameters, in a bid to rival leading AI models from the likes of OpenAI, Anthropic, and Google. Spearheaded by ex-Google DeepMind founder Mustafa Suleyman, who recently joined Microsoft post the acquisition of AI startup Inflection, MAI-1 represents a significant venture for the tech giant. Although not directly derived from Inflection's models, technological expertise and training methodologies from the acquired company might inform the development of MAI-1. This move signifies Microsoft's ambition to broaden its AI capabilities, with MAI-1 complementing smaller, on-device models like the Phi family. As Microsoft diversifies its AI portfolio, questions arise regarding the future of its partnership with OpenAI — is it evolving or heading towards a potential split?

OpenAI Enhances Transparency of AI-Generated Content

OpenAI is making strides towards enhancing transparency in AI-generated content by joining the Coalition for Content Provenance and Authenticity (C2PA) steering committee and incorporating the open standard's metadata into its generative AI models. This move aims to certify digital content origins, whether entirely AI-created, AI-edited, or traditional. OpenAI has already started integrating C2PA metadata into images generated by its latest DALL-E 3 model in ChatGPT and the OpenAI API, with plans to extend this to its upcoming video generation model, Sora. Amidst concerns about AI-generated content misleading voters ahead of major elections, authenticating AI-created media could aid in combatting deepfakes and disinformation campaigns. OpenAI is also developing tamper-resistant watermarking for audio and image detection classifiers to identify AI-generated visuals, aiming for increased adoption of provenance standards to ensure metadata accompanies content throughout its lifecycle, thus enhancing digital content authenticity practices. Additionally, OpenAI is collaborating with Microsoft to launch a $2 million societal resilience fund supporting AI education and understanding, emphasizing the necessity for collective action in enabling content authenticity.

Source: Pexels

Ukraine's Fresh AI Representative

Ukraine’s Ministry of Foreign Affairs has unveiled 'Victoriya Shi', an AI-generated avatar serving as the department’s digital spokesperson, marking a groundbreaking development in diplomatic practices worldwide. Victoriya Shi will deliver pre-prepared official statements on behalf of the ministry, with her visual avatar entirely generated by AI and her voice modeled after Rosalie Nombre, a notable Ukrainian singer and reality TV personality. In a proactive measure against deepfakes, each of Victoriya's statements will feature a unique QR code linking to the official text version on the ministry's website. This innovative initiative positions Ukraine as a pioneer in employing AI technology in government communication, hinting at a future where AI-driven diplomats could handle more complex tasks and communications beyond visual representation alone.

Newsletter

📬 Receive our amazing posts straight to your inbox. Get the latest news, company insights, and Niural updates.

Thank you! Your message has been received!
Oops! Something went wrong. Please fill in the required fields and try again.

Q&Ai

How is VISA employing generative AI to combat account fraud attacks?

VISA is leveraging generative AI to combat account fraud attacks, a pressing issue in the realm of digital transactions. Attackers are increasingly utilizing automation, including botnets and scripts, to carry out fraudulent card-not-present (CNP) transactions, resulting in significant financial losses. These enumeration attacks pose a serious threat due to their speed and scale, targeting vulnerabilities in e-commerce platforms and digital payment systems. To address this challenge, VISA has introduced Visa Account Attack Intelligence (VAAI), integrating breach, cyber, and payment intelligence to detect and prevent fraud. Now, VISA is enhancing its fraud prevention capabilities with a new genAI-powered VAAI score, which assesses the risk of fraudulent transactions in real-time, enabling issuers to make faster, more accurate decisions. This innovative solution, developed from analysis of billions of transactions, promises to reduce false positives and enhance network security significantly. As VISA continues to invest in AI and machine learning technologies, the battle against fraud in digital transactions enters a new era of efficiency and effectiveness.

 How to build AI apps in hours?

Navigator empowers teams to swiftly and efficiently construct AI-powered solutions and seamlessly deploy them into production, irrespective of machine learning resources. This platform facilitates the acceleration of AI projects by enabling the training and tuning of models within minutes to hours while ensuring the maintenance of data ownership and confidentiality. With Navigator, users can deploy their solutions effortlessly to local edge devices or cloud-based platforms and have the flexibility to build using either full code or a drag-and-drop interface, catering to varying levels of expertise.

Tools

💬 TLM Playground is Clanlab's new trustworthy ‘Hallucinationless‘ Language Model

🧑‍💼 Braintrust AIR is the first end-to-end AI recruiter

n8n connects and automates anything via an open, fair-code model

🎥 Wizardshot automatically creates step-by-step tutorials with AI

🗣️ Perplexity Voice asks questions and hear answers, available on iOS for Pro users

Follow us on Twitter and LinkedIn for more content on artificial intelligence, global payments, and compliance. Learn more about how Niural uses AI for global payments and team management to care for your company's most valuable resource: your people.

See you next week!

Request a demo