AI is a technology not a product
Framed around the claim that “AI is a technology, not a product,” commenters debate whether companies like Apple should treat large language models as invisible infrastructure that quietly improves existing experiences (e.g., Siri, search, spam filtering) rather than standalone “AI products.” Many argue current AI hype ignores real user needs—phones still dominate, basic assistants remain unreliable, and many people don’t actually want agents to automate everyday “life tasks” like trip planning or grocery lists—while others see clear value in better translation, accessibility, and task delegation once models are cheap, local, and tightly integrated. Underneath the back-and-forth is a broader question: will AI reshape interfaces and devices, or just become another commodity layer akin to microprocessors or TCP/IP that only matters in how it’s built into concrete products.
AI as Technology vs. Product
- Many see current “AI products” as misframed; models are likened to microprocessors, TCP, or Dropbox-style sync: foundational tech, not end-user products.
- Consensus from several comments: real value comes when AI is invisibly embedded into concrete workflows, not presented as “an AI app” or brand.
- Some expect AI models to become commodity infrastructure (like Linux), with differentiation at the hardware, UX, and integration layers.
Apple, Siri, and “Working Backwards”
- Repeated theme: Apple should treat AI as a way to fix Siri and system UX, not as a standalone AI brand.
- Desired capabilities: natural-language calendar creation, robust app control (“play this podcast in this app”), smarter Shortcuts, better speech recognition for non‑US accents, and unified retrieval of context (e.g., “what’s tonight’s dinner about?”).
- Frustration that Siri remains brittle and unreliable even for basics like timers, lights, and reminders.
- Some argue Apple’s slow roll is rational: phones remain central, they can buy model access, and chasing frontier models is costly and risky.
Real-World Usefulness of LLMs
- Debate over whether LLMs materially improve non‑coders’ lives.
- Pro side: cheap, always‑available “good enough” expertise; easier website/content creation; translation; search-like help.
- Skeptical side: many of these things existed (search, Google Translate, Squarespace); hallucinations and misinformation may outweigh benefits; some claim LLMs “aren’t even useful for coding.”
Agents, Automation, and UX
- Strong split on AI “agents” that auto‑order rides, plan life, etc.
- Critics see this as infantilizing, dystopian, or solving non‑problems; many people actually enjoy planning and everyday tasks.
- Supporters compare it to having a personal assistant, especially valuable amid dark patterns, complex travel, or accessibility needs.
- Voice is viewed as powerful for narrow tasks (alarms, simple queries, accessibility) but poor for dense information and privacy; several argue for more deliberate, limited use of voice UIs.
Devices and Form Factors
- Some insist the phone form factor will dominate for years; others argue long‑term convergence toward watches or glasses with AI-centered interaction.
- Differing views on “always-on, fully integrated” wearables: appealing to some, intrusive to others who value being able to put the phone away.
Local Models, Ecosystem, and Trust
- Interest in small local models combined with web search to reduce dependence on corporate clouds and bias.
- Some praise other platforms for already shipping “AI as feature” (better spam detection, visual search, call handling).
- Concerns raised about attention abuse, dark patterns, and “slop” content; one vision is an “anti‑AI” layer that flags or filters low‑quality AI‑generated material.
Meta: Perceptions of the Blogger
- A subthread criticizes the blog author’s political and ethnic commentary in other contexts, questioning their judgment and bias.