AI in Marketing

Google's latest AI image model, Nano Banana Pro, officially known as Gemini 3 Pro Image, is making waves for its impressive capabilities. Developers and enterprise users are calling it "absolutely bonkers" due to its ability to generate high-quality, multimodal images with precision and accuracy. Unlike previous models aimed at casual users, this one is designed for structured workflows, offering high-resolution outputs, multilingual accuracy, and real-time knowledge grounding. It's integrated across Google's AI stack, including Gemini API, Vertex AI, and Workspace apps, making it a versatile tool for enterprise-scale automation. Benchmarks show it outperforms competitors in visual quality and text rendering accuracy, making it a valuable asset for creating infographics, medical illustrations, and more. The model supports resolutions up to 4K and includes advanced controls over camera angles and lighting. Pricing is competitive, with tiered costs based on resolution and usage. Each image includes SynthID, Google's digital watermarking system, ensuring compliance and provenance. While early reactions are overwhelmingly positive, some tests reveal its limitations in logic-heavy tasks. Overall, Nano Banana Pro is a significant addition to Google's AI ecosystem, offering a new level of visual reasoning and creativity for enterprise applications.

From Nano Banana Pro to Mixup, AI creativity takes a social turn

Mixup, a new iOS app from the team behind Rooms at Things, Inc., offers a creative twist on AI photo editing. Users can generate unique images by filling in "recipes," which are Mad Lib-style prompts for photos, text, or doodles. The app, built on Google's Nano Banana, allows users to transform sketches into Renaissance paintings or reimagine pets in costumes. Mixup's shareable recipes make it a social experience, letting users publish their creations or keep them private. The app addresses the unpredictability of AI-generated images by showing both the prompt and resulting photo, offering users more control. Users can upload personal photos, enabling friends to create images with their likeness, though privacy settings allow for control over this feature. Mixup uses OpenAI technology for moderation, alongside Google's controls to restrict inappropriate content. Launching globally on November 21, the app is free with 100 credits, with subscription options available. An invite is required for access, and a web or Android version may follow if successful.

Turning to group chats, Mixup's collaborative spirit finds a new platform

OpenAI has globally launched group chats for ChatGPT users across all plans, following a successful pilot in Japan and New Zealand. This new feature allows up to 20 participants to collaborate in a shared conversation, transforming ChatGPT from a one-on-one assistant into a collaborative space. Users can coordinate trips, co-write documents, and work through research, with ChatGPT assisting in searching, summarizing, and comparing options. Each participant's settings and memory remain private, and adding someone to a chat creates a new conversation. Users can tag ChatGPT for responses, and the AI can react with emojis. This development is part of OpenAI's broader strategy to evolve ChatGPT into a more interactive and social platform.

From ChatGPT's social features to peer corrections, online interactions evolve

A study by researchers from the University of Rochester and others reveals that peer-driven corrections on social media can effectively reduce misinformation. The research focused on X's Community Notes, where users flag misleading posts. When a community note appears publicly, authors are 32% more likely to delete their posts compared to private notes. This peer feedback mechanism, which relies on diverse user input, encourages authors to retract misleading content voluntarily, preserving free speech while curbing misinformation. The study analyzed 264,600 posts, noting that public corrections not only increased deletions but also sped up the process. Verified users, concerned about their reputation, were particularly quick to act, highlighting the role of social dynamics in promoting online accuracy.

In education, from peer feedback to AI-driven courses, challenges arise

Students at the University of Staffordshire are expressing frustration after discovering their coding course was largely taught using AI-generated materials. James and Owen, among 41 students in a government-funded apprenticeship program, noticed AI-generated slides and voiceovers, leading to doubts about the course's value. Despite raising concerns, the university continues using AI, justifying it with a policy statement on leveraging AI in teaching. Students feel this contradicts the university's strict policies against AI use in student work. James, midway through his career, feels stuck, unable to restart. Owen, seeking foundational knowledge, finds the course lacking substance. The university claims academic standards are upheld, but students remain dissatisfied, feeling their time was wasted. This case highlights growing tensions as more universities incorporate AI into education, with students often finding it demoralizing rather than beneficial. Despite a non-AI lecturer's involvement in the final session, students believe it was insufficient to address their concerns.

From AI in education to PlushDaemon's exploits, digital vulnerabilities persist

A Chinese state-aligned group, "PlushDaemon," has been quietly exploiting software updates to spy on Chinese organizations for over five years. By infecting routers and similar devices, they reroute legitimate update requests to their own servers, delivering malicious software instead. Their malware, "EdgeStepper," targets popular Chinese applications like Tencent QQ and WPS Office. Once installed, it leads to the download of "SlowStepper," a backdoor that steals sensitive data. Despite its long activity, PlushDaemon has largely evaded detection, raising questions about its motives, especially since it primarily targets Chinese entities. ESET researchers suggest focusing on securing network devices to prevent initial infections, emphasizing the importance of addressing vulnerabilities and strengthening credentials to thwart such attacks.

On a broader scale, from PlushDaemon to AI market concerns, security and stability are key

Alphabet CEO Sundar Pichai warns that no company, including Google, is immune if the AI bubble bursts, as he told the BBC. He acknowledges the current AI investment wave as extraordinary but notes market irrationality reminiscent of the dotcom era. Despite Alphabet's 46% share surge this year, driven by competition with OpenAI, concerns about AI valuations are affecting broader markets. In the UK, Alphabet has committed £5 billion for AI infrastructure, including a new data center and DeepMind investments. Pichai also highlights the significant energy demands of AI, which may delay Alphabet's net-zero goals. Meanwhile, Google plans to train AI models in Britain, aligning with the country's ambition to become a leading AI superpower.