AI News – June 8, 2023

EU Calls for AI-Generated Content Labeling to Combat Disinformation

The European Union is urging signatories of its Code of Practice on Online Disinformation to label deepfakes and other AI-generated content. The EU’s values and transparency commissioner, Vera Jourova, has called for technology that can recognize and clearly label AI content for users. The aim is to prevent the misuse of AI technologies for the creation and dissemination of disinformation. The EU sees two main areas for discussion: services integrating generative AI, such as Microsoft’s New Bing or Google’s Bard AI-augmented search services, should build in safeguards against misuse for disinformation, and signatories with services that could disseminate AI-generated disinformation should implement technology to recognize and label such content. The EU is pushing for immediate implementation of these measures. ¹

Automattic Introduces AI Writing Assistant for WordPress

Automattic, the company behind WordPress.com, has launched an AI assistant for the popular content management system. The AI assistant integrates with WordPress.com and all Jetpack-powered sites, providing users with a tool that generates text based on prompts in natural language. The AI assistant can create structured lists and tables, adjust the tonality of a post, create summaries, and suggest titles. It supports 12 languages and offers translation capabilities, allowing users to write in their native language and translate it to English or other languages. The AI assistant also provides enhanced spelling and grammar correction features. After a free trial of 20 requests, users can access the feature for $10 per month. ²

DeepMind Utilizes Game-Playing AIs for Code and Infrastructure Optimization

DeepMind’s Alpha series of AIs, known for their prowess in games like Go, Chess, and Shogi, are being repurposed to tackle tasks beyond gaming. AlphaZero, for instance, has been applied to Google’s Borg system, which manages task assignment at data centers. By identifying patterns in data center usage and incoming tasks, AlphaZero was able to reduce the amount of underused hardware by up to 19%. Similarly, MuZero was used to analyze YouTube streams, resulting in a 4% reduction in video bitrate. Another variant, AlphaDev, improved sorting algorithms and reduced the load by 30% in a specific hashing function. These applications demonstrate the potential of game-playing AIs in optimizing code and infrastructure, indicating a promising level of flexibility and robustness in AI systems. ³

Google’s Bard AI Chatbot Shows Improvement in Math and Programming

Google’s AI-powered chatbot, Bard, is reportedly improving in tasks involving logic and reasoning, particularly in math and coding. According to a recent blog post by Google, Bard’s progress is attributed to a technique called “implicit code execution,” which allows Bard to write and execute its own code. This new approach enables Bard to identify prompts that might benefit from logical code, write the code internally, test it, and use the result to generate a more accurate response. Google’s internal benchmarking indicates that Bard’s responses to computation-based word and math problems have improved by 30% compared to the previous release. However, Google acknowledges that Bard may not always generate correct responses. ⁴

Adobe Introduces Firefly Image Generator and Enhanced Sensei GenAI to Enterprises

Adobe has announced the introduction of its AI-powered Firefly generative image generator to enterprise customers, allowing them to customize the model with their own branded assets. In addition, Adobe is also bringing its Adobe Express design app to enterprise users, who will have access to Firefly. Adobe’s Firefly has been integrated into Photoshop and has been used over 150 million times. The company is banking on the fact that Firefly produces commercially safe images, which are trained on images from Adobe’s stock imagery marketplace, openly licensed images, and public domain content. Alongside Firefly, Adobe has also launched several other generative AI-powered services as part of its Sensei GenAI platform, which includes a generative AI-based marketing copy generator. ⁵

Google Cloud Enhances AI Portfolio with New Tools, Programs, and Partnerships

Google Cloud is expanding its foray into generative AI with a host of new tools, services, and programs designed to assist businesses in adopting and utilizing this emerging technology. The initiative includes increased access to Google’s generative AI products like Vertex AI and Duet AI, new educational programs and consulting services, blueprints for specific use cases, and expanded partnerships with companies such as DataStax, Neo4j, Twilio, and Typeface. Google’s Vertex AI platform is now more widely available, offering access to over 60 machine learning models that can generate images, translate languages, summarize text, and more. Google Cloud Consulting has also launched several free learning programs covering generative AI for business executives, software developers, and Google Cloud customers. ⁶

Sources