Cognitive warmup. The closer we get to the Apple iPhone keynote in September, the more you’ll begin to hear these nuggets of expected new features. Some will be true, some won’t, that is just how things are. Apple’s AI push, as part of the Apple Intelligence suite is expected to evolve significantly with iOS 27. There is some murmur that Clean Up, which is already part of Photos, will add more to its arsenal of capabilities—extend a photo beyond the original frame, enhance elements such as lighting and quality, as well as reframing a photo to shift focus and perspective later. John Ternus will be CEO by the time the iPhone keynote comes around, and don’t expect him to take the AI push lightly. The Google Gemini partnership only lends credence to an expected trajectory.
20 years and staying relevant
A few days ago, Google Translate turned 20. This is another Google tool that remains incredibly relevant, withstanding the test of time (and a changing ecosystem landscape as is always the case). To mark this milestone, Google has added something called pronunciation practice, for now on the Translate app on Android. This goes live in two regions, that is US and India, and in three languages (English, Hindi and Spanish), at this time.
The key here is, this builds on the “ask and understand” capabilities, and pronunciation practice will use AI to analyse a user’s speech and provide feedback on best pronunciation. Google also took this moment to confirm that Translate has been using statistical machine learning since 2006, which has increasingly improved over time for fluency and natural translations. Google Translate, in this moment, supports 250 languages and more than 60,000 potential language pairs. In terms of usage metrics, across Translate, Search, Lens and Circle to Search, more than 1 trillion words are being translated every month by users worldwide.
Meta’s humanoid machine idea
Meta is acquiring the US-based AI robot startup, Assured Robot Intelligence (ARI), as it wants to build the AI play with machines for high-value labour use-cases. And it has, while at it, given us a new terminology to work with for the next few months (and I’m sure, wrap millions of dollars worth of industry-wide funding too, while at it)—Physical AGI. ARI co-founder Xiaolong Wang, in a post on X, says “ When we started ARI one year ago, our mission was clear: achieve physical AGI. Through deep customer engagements and real-world deployments, it became clear to us that serving the massive opportunity ahead requires training a truly general-purpose physical agent”. We still don’t have a definition of progress for the Artificial General Intelligence (AGI) that has been the favourite buzzword of AI bros over the past year (tell me I’m wrong, and we have a working definition), but great, let us go along with Physical AGI.
The way this is supposed to work is, ARI will provide the expertise with models and frontier capabilities essential for humanoid and robot management, while Meta is likely to focus on the hardware side of things. This is where Meta wants to be seen competing (or at least working in the same space) as Amazon and Tesla already do. Reminds me of something Meta’s CTO Andrew Bosworth said last year, about creating software that other companies can license.
Lawyers, Microsoft and trust
Microsoft wants lawyers and anyone in the legal profession, to trust its AI inside Word. It is a new Legal agent, potentially doing some of the analysis, summarisation and highlighting. “Instead of relying on general AI models to interpret commands, the agent follows structured workflows shaped by real legal practice, managing clearly defined, repeatable tasks like reviewing contracts clause by clause against a playbook,” explains Sumit Chauhan, corporate vice president of Microsoft’s Office Product Group.
The Legal Agent, Microsoft says, understands complex legal documents, can track changes from prior proposals for new versions while maintaining historical log, and can compare versions for spot risks and obligations. Coincidentally, Microsoft had acquired Robin AI a few months ago—a startup that was building an AI based contract review system.
Neural Dispatch is your weekly guide to the rapidly evolving landscape of AI. Want this newsletter delivered to your inbox? Subscribe here.
