Apple’s ambitious plans to overhaul Siri with cutting-edge AI capabilities have hit a significant delay, with the next-generation “LLM Siri” now slated for release in 2026, according to Bloomberg´s Mark Gurman.
Designed to rival competitors like Google’s Gemini Live and OpenAI’s Advanced Voice Mode, the new Siri will feature Apple´s own large language models (LLMs) enabling natural conversations and multi-step task execution. However, internal documents reveal the project is far from completion, with Siri trailing behind competitors in both accuracy and functionality.
This delay comes as Salesforce CEO Marc Benioff recently lauded Google’s Gemini Live as the “future of consumer AI,” further emphasizing Siri’s struggles to maintain relevance in a rapidly evolving market.
Siri’s AI Journey: Incremental Updates and Long-Term Vision
Apple has been working to keep Siri competitive through incremental updates. The upcoming iOS 18.2 update, currently in beta and expected in December 2024, will integrate third-party AI tools like OpenAI’s ChatGPT and Google’s Gemini.
These tools will temporarily enhance Siri’s capabilities, allowing for features such as real-time email drafting and contextual photo editing. However, these additions highlight Apple’s dependence on external AI solutions while its proprietary LLM Siri remains in development.
At WWDC 2024, Apple unveiled its “Apple Intelligence”, also promising visual and functional updates to Siri. These include a glowing interface, the ability to type queries, and improved contextual awareness. Yet, these changes are primarily cosmetic, with Siri still unable to match the conversational depth and flexibility offered by competitors like Gemini Live and OpenAI’s Advanced Voice Mode.
Internal Struggles: Siri’s Lag Behind Competitors
Leaked internal evaluations from October 2024 revealed that Siri’s accuracy trails OpenAI’s ChatGPT by 25%, with the latter successfully answering 30% more queries. These findings underscore the significant challenges Apple faces in developing AI systems capable of competing with the likes of Google and OpenAI.
Apple’s focus on on-device processing—a cornerstone of its privacy-first strategy—has limited Siri’s ability to leverage the vast datasets required for advanced machine learning models. While this approach safeguards user data, it also hampers Siri’s performance compared to cloud-based systems.
Apple researchers, including Samy Bengio, have developed tools like GSM-Symbolic to test AI reasoning capabilities, revealing limitations in existing models. These tests demonstrate that even small adjustments to query phrasing can significantly confuse AI systems, highlighting the complexity of achieving robust conversational AI.
Google’s Gemini Live Sets the Pace
Google’s Gemini Live debuted at the Pixel 9 event in August 2024 and has quickly become a benchmark for voice-based AI assistants. Offering zero-latency responses and hands-free multitasking, Gemini Live supports uninterrupted conversations even when devices are locked. Its rapid adoption, including availability on iOS, has positioned it as a direct competitor to Siri.
Salesforce CEO Marc Benioff recently praised Gemini Live’s real-time capabilities, while he didn´t hold back about Microsoft’s Copilot, which he criticized earlier as “Clippy 2.0.” Despite its strengths, Gemini Live faces its own challenges, including limited voice customization and the occasional “hallucinations”—where the AI provides inaccurate or outdated information.
Amazon’s Alexa: Subscription Models and Claude AI
Amazon’s voice assistant Alexa is also undergoing significant changes under “Project Banyan.” Delayed to 2025, the project aims to introduce more advanced conversational capabilities, the $4 billion partnership with Anthropic for their Claude AI will surely play a big part in this.
The upcoming Alexa Plus will feature a subscription model priced at $5–$10 per month, sparking criticism over whether users will pay for features increasingly seen as standard in competitors like Siri and Gemini Live.
|
Related: |
OpenAI and Microsoft: Competing Approaches
Meanwhile, OpenAI continues to lead in voice AI with its Advanced Voice Mode, which expanded to Europe in October 2024. Known for its sub-320-millisecond response times and customizable intonations, the tool outpaces Siri and Gemini Live in conversational fluidity.
Meanwhile, Microsoft’s Copilot AI focuses on productivity rather than voice interaction, drawing criticism from Benioff for failing to meet the modern standards of conversational AI. Well, conversational AI was never Microsoft´s thing in the first place. Those who remember and actually tried to use the long gone Cortana assistant can attest to that. It´s now all Copilots, running technology provided primarily by OpenAI – not a dumb move.
Last Updated on December 10, 2024 7:36 pm CET
