Apple’s New Chatbot Rivals ChatGPT but Struggles With Accuracy

Fast Summary

• Apple’s AI-focused event, WWDC 2025, is expected to prioritize basic updates amid mounting criticism of its AI offerings.
• Apple Intelligence, the company’s lightweight AI initiative launched in June 2024, uses LLMs with parameters around three billion but faces important limitations.
• Larger and more advanced models (7 billion, 33 billion, and 150 billion parameters) are internally active but not ready for public release due to issues like factual inaccuracies (“hallucinations”).
• Internal tests using a tool called Playground show that the most advanced model rivals chatgpt’s quality but suffers from output reliability concerns.
• Diverging opinions among Apple executives regarding when-or whether-to unveil newer chatbot technology have created internal debate about managing public perception versus avoiding premature rollouts.

Image: !Siri orb logo inside apple intelligence logo against a purple red and orange background

Indian Opinion Analysis
Apple’s cautious approach in developing generative AI models highlights broader concerns within the global tech industry about balancing innovation with ethical risks such as misinformation through inaccurate outputs or “hallucinations.” For India-rapidly advancing as a technological hub-this news underscores how critical it is indeed for emerging startups to tackle similar challenges responsibly while adopting scalable solutions for generative AI development.

Given India’s growing digital economy and investment in artificial intelligence initiatives, Apple’s deliberation may encourage stricter scrutiny across sectors on deploying trustworthy algorithms in consumer-facing products like chatbots. However, Apple’s decision on whether to release its advanced models at WWDC could serve as either a benchmark or cautionary tale-a choice that Indian companies would be wise to monitor closely given increasing domestic demand for robust conversational AI technologies.

Read More