- The Upload
- Posts
- Voice AI Makes a Breakthrough
Voice AI Makes a Breakthrough
New Model Crosses 'Uncanny Valley' While Siri Overhaul Faces Major Delays
Mark Chepelyuk
March 03, 2025
Voice AI reaches a watershed moment with Sesame's model achieving natural, emotionally aware conversations that are "undetectably human." Meanwhile, Apple's Siri overhaul faces significant delays with a complete revamp potentially not arriving until 2027, creating a widening capability gap against competitors like Amazon's Alexa+.
Here’s what you need to know about AI today:
Sesame's voice AI crosses the "uncanny valley"
Apple's Siri overhaul delayed until 2027
Claude 3.7 sparks coding debate among developers
OpenAI confirms Sora integration with ChatGPT
February's major AI model releases compared
Today’s Deep Dive:
Image Source: Sesame
🗣️ Voice AI's Breakthrough Moment
Oculus co-founder's startup Sesame launches AI voice tech with unprecedented natural speech and emotional awareness.
Key Details:
Considers conversation context in real-time
Adjusts tone and rhythm based on emotional content
Incorporates natural pauses and pacing
Maintains threads when interrupted
Also developing AI glasses with integrated voice tech
Why It Matters: After years of stilted, robotic voice assistants, we're witnessing a fundamental shift in how humans and AI can verbally interact. This breakthrough could accelerate the transition from screen-based interactions to more natural voice-based computing, potentially reshaping how we engage with technology in our daily lives.
What This Means For You: Voice interactions with AI are about to become dramatically more natural and useful. As this technology matures and becomes integrated into consumer products, expect voice interfaces to finally deliver on their long-promised potential.
Image Source: Apple
🍎 Apple's AI Emergency
Apple's comprehensive Siri revamp reportedly delayed until 2027.
Key Details:
Current Siri uses fragmented architecture
Planned integration behind schedule
Users not adopting Apple Intelligence features
Division facing talent poaching and leadership changes
Struggling to secure necessary AI chips
Why It Matters: Apple's traditional "perfection over promptness" approach is becoming a liability in the rapidly evolving AI landscape. With Amazon's Alexa+ and Google's Gemini making significant advances, a 2027 timeline for a competitive Siri would leave Apple essentially sidelined in the voice AI race for years.
What This Means For You: If you're in the Apple ecosystem expecting significant Siri improvements soon, you may need to adjust your expectations or consider alternative assistants. This delay could also impact Apple's hardware strategy and market position as AI capabilities become increasingly central to consumer technology choices.
Image Source: Anthropic
👨💻 Claude's Coding Controversy
Power users debate Claude 3.7's coding abilities, with some reverting to 3.5.
Key Details:
Some developers reporting Claude 3.7 overdesigns code
400-line scripts expanding to 1,100 lines
Reports of instruction following issues
Better for autonomous problem-solving than collaboration
Success requires detailed initial prompts and clear constraints
Why It Matters: This debate reveals an important shift in AI development: as models become more sophisticated, they may transition from collaborative partners to autonomous problem-solvers requiring different interaction approaches. The mixed reception also demonstrates that newer isn't always better for specific use cases.
What This Means For You: If you're using Claude for coding, you might need to adapt your workflow based on your specific needs. For collaborative coding assistance, 3.5 might still be preferable, while 3.7 excels with comprehensive instructions and clear boundaries for autonomous tasks.
🎬 Sora Expands Its Reach
OpenAI confirms plans to integrate Sora video generation directly into ChatGPT.
Key Details:
Revealed during "Sora Global Office Hours" on Discord
ChatGPT version likely to have limited functionality
Dedicated mobile app also in development
Sora-powered image generator to potentially replace DALL-E 3
Faster "Sora Turbo" model in the works
Why It Matters: By integrating Sora into ChatGPT, OpenAI aims to make video generation more accessible and part of a unified creative workflow. However, the competition in AI video has intensified with Google's Veo 2, Alibaba's Wan 2.1, and other models emerging with impressive capabilities.
What This Means For You: If you're already using ChatGPT, you'll soon be able to generate videos without switching platforms. However, the limited functionality compared to the standalone Sora experience suggests that serious video creators may still need specialized tools.
Image Source: Getty
📊 February's AI Roundup
February saw major AI model releases from leading companies with distinct strengths and weaknesses.
Key Details:
GPT-4.5 excels at conversation but struggles with complex reasoning
Claude 3.7 Sonnet shows strong coding performance but mixed developer reception
Grok 3 offers real-time information access but lacks content safeguards
DeepSeek R1 remains influential despite being open-source
Each model serves different use cases and workflows
Why It Matters: The diversity of these models highlights how AI is becoming increasingly specialized rather than one-size-fits-all. The varied strengths and weaknesses demonstrate that AI development is branching into distinct approaches rather than converging on a single best solution.
What This Means For You: Choosing the right AI tool now requires more consideration of your specific needs and use cases. For creative writing, GPT-4.5 might be ideal; for coding, Claude 3.7 or 3.5 depending on your workflow; for unfiltered responses, Grok 3; and for developers wanting to understand the underlying technology, DeepSeek R1.
🛠️ Trending AI Tools
🎥 Pika 2.2: 10s video generations for longer, more dynamic clips
🤗 GPT-4.5: OpenAI’s latest and largest model
🎬 Wan 2.1: Alibaba's open-source video generation suite
💭 Inception Labs: The first diffusion large language models