After years of false starts and incremental updates, Apple's AI-enhanced Siri has finally arrived in public beta, and its most revolutionary feature may be what it does not do: it does not try to become your friend. This marks a distinct shift in the landscape of virtual assistants, where competitors like ChatGPT, Claude, and Gemini often strive to appear human and build emotional bonds. Siri, in contrast, remains purely functional—a tool that gets the job done and then steps aside.

The new Siri is currently available to beta testers as part of iOS 27 and macOS 27, with a wide release expected this fall. Early testers, including this writer, have been struck by the assistant's business-like demeanor. It is not curt or blunt, but it also does not shower users with praise, extend conversations unnecessarily, or attempt to mimic human empathy. After completing a task, the most effusive response you will hear is a simple "Here you go."

What the new Siri can do

Despite its understated personality, the revamped Siri is genuinely powerful. One of the standout capabilities is email management. The assistant can sift through inboxes, organize messages by relevance or date, and even draft replies based on context. It can also brainstorm ideas, create Notes files on the fly, and file them into designated folders without additional prompting. This is the kind of practical, everyday utility that users have long wanted from a voice assistant.

Another impressive example involves retrieving information from multiple sources. When asked about a child's field trip, Siri was able to automatically pull details from email messages and text conversations, providing the exact time, location, and teacher's contact number. This cross-app intelligence—combining data from Mail, Messages, and Calendar—demonstrates a level of integration that third-party assistants simply cannot match due to platform limitations. It is the kind of seamless, context-aware help that makes users wonder why it took so long to arrive.

Perhaps most importantly, when the task is complete, Siri disappears. There is no follow-up chatter, no attempt to keep the user engaged, and no personality-driven interaction. The assistant is happy to help, but it is not a companion, and it never tries to be the center of attention.

A deliberate contrast with competitors

Apple's approach stands in stark contrast to the direction taken by OpenAI, Anthropic, and Google. ChatGPT, especially in its latest iterations, often feels like it is trying to bond with users. It uses conversational fillers, offers unsolicited encouragement, and sometimes extends dialogues beyond what is necessary. A recent report indicated that OpenAI's rumored smart speaker is designed to "feel like a companion" and "connect on a humanlike level with users." Similarly, Claude and Gemini frequently use language that mimics human warmth, which can be helpful in some contexts but also feels intrusive to users who simply want answers or actions.

Siri's refusal to engage in emotional labor is not a lack of intelligence—it is a design choice. Apple seems to understand that for many tasks, the best assistant is one that stays out of the way. This philosophy also aligns with Apple's long-standing emphasis on privacy. By not building a personality profile or tracking emotional states, Siri avoids the data collection that often accompanies more human-like AI. Users can ask sensitive questions or complete personal tasks without feeling like they are being psychoanalyzed.

This utility-first approach may prove more sustainable in the long run. As AI assistants become ubiquitous, the novelty of talking to a computer that acts like a human may wear off. What will remain is the need for reliable, efficient, and unobtrusive help. Siri's quiet revolution is in recognizing that the best assistant is the one that does not call attention to itself.

More in AI this week

• The latest iteration of ChatGPT, powered by a more advanced model, can now write code, control a computer's desktop interface, browse the internet, and even publish websites directly from the chat window. This represents a significant expansion of capabilities beyond simple conversation.
• One test involved asking ChatGPT to play a game of chess and write a journal entry about the experience. The assistant successfully managed both tasks, demonstrating its ability to handle sequential, multi-step instructions across different domains.
• OpenAI also released a dedicated hardware device—a compact keyboard with programmable hotkeys for its Codex coding tool. Priced at $230, the device is aimed at developers who want fast access to AI-assisted coding commands without leaving their keyboard.
• Meta faced backlash and quickly withdrew a controversial feature that allowed users to generate deepfake images of public Instagram photos. The feature raised significant privacy and consent concerns, prompting the company to pull it within hours.
• Anthropic and OpenAI reset usage limits for their most advanced models, giving users more flexibility to experiment with Claude and GPT-5.6 Sol. This expansion of access allows developers and power users to test the limits of the latest AI without worrying about hitting caps mid-session.

Prompt of the week: The “pencil and paper” tip

When using advanced coding assistants like Claude Code, it is common for the AI to immediately jump into writing code as soon as it understands the problem. This can be counterproductive if you simply want to discuss a bug or architecture before making changes. A useful technique is to add the phrase "Stay in pencil and paper mode" at the end of your prompt. This instructs the assistant to explain its reasoning, provide analysis, and outline possible solutions without actually modifying any files or writing executable code.

The phrase works beyond coding. It can be applied to any situation where you want an AI to think aloud before taking action. For instance, you can ask an LLM to outline a meal plan, brainstorm product features, or evaluate pros and cons—all without committing to a final output. This keeps the conversation focused on strategy and understanding, rather than premature execution. It is a simple but effective way to maintain control over the assistant's pace and output.

Source: PCWorld News