It’s quite fascinating to note how Apple has transitioned over the past 14 years from a leader in intelligent assistants to being significantly behind in the game.
We’ve seen the evolution from the innovative Siri introduced in 2011 to the rather inadequate Apple Intelligence of 2025…
14 Years of Siri Evolution
I vividly recall the keynote for the iPhone 4S, where Siri was the standout feature that compelled me to get one immediately.
Although Apple did not originally create Siri—having first launched as a third-party app—it was under Steve Jobs’s leadership that Apple acquired the technology and integrated it into the iPhone, elevating intelligent assistants to a crucial expectation in modern smartphones.
A span of fourteen years should have enabled Apple to transform Siri into an exceptionally powerful assistant, yet that transformation remains unrealized. Back in 2015, I shared my vision for the capabilities I wished to see, including enhanced interaction with apps. Remarkably, it took a whole decade for Apple to start implementing this feature!
What’s even more astonishing is that in 2018, I compiled a list of basic functionalities that Siri still lacked, and several of those deficiencies remain unaddressed today!
Currently, Siri resembles a Lada, while ChatGPT and its counterparts have become the Mercedes of the AI world.
Nonetheless, Apple does have a few explanations for its struggle to match the performance of contemporary LLMs.
Reliability: A Key Issue
First, let’s discuss reliability.
Companies like OpenAI and Google took a “move fast and break things” approach to AI development. During the early days of ChatGPT and Google Bard, I pointed out instances where their outputs were hilariously incorrect. For example, ChatGPT erroneously generated fictitious references for a scientific paper, and Google Bard misanswered a question during a live demo meant to showcase its intelligence.
I stressed that Siri’s verbal responses amplify the dangers of such errors.
When it comes to the risk of misinformation, receiving a confidently incorrect answer can be more perilous than not knowing something critical at all.
While Google search results can present conventional responses next to a chat window with an accompanying disclaimer regarding possible inaccuracies, Siri’s design is predicated on delivering spoken answers to verbal inquiries. An even more frustrating scenario than Siri merely stating, “Here’s what I found online,” would be “Here’s a lengthy answer you need to listen to, followed by a note that it might not be correct, so you should look online for confirmation.”
Privacy Concerns
The other significant factor is privacy. Siri has always adhered to two foundational principles that prioritize user privacy:
- Processing on-device whenever possible
- Anonymized requests when relying on Apple servers
On-device processing inherently lacks the sophistication of operations managed by robust data center machines, and ensuring that a Siri server remains unaware of user identity restricts its ability to be as knowledgeable as a Google server, which has insight into a user’s search history and personal data.
I previously argued that while the anticipation for an advanced LLM version of Siri is frustrating, the long-term privacy benefits will ultimately justify the wait.
A wealth of personal data exists on our devices that could empower Siri to become a genuinely intelligent assistant—information stored in Apple Calendar, Contacts, Health, Mail, Maps, Messages, Wallet, among others. Users will also have the power to grant Siri access to particular third-party apps, right on their devices.
Once Siri can access these applications, it can evolve into a capable competitor relative to other systems, while still preserving user privacy.
The AI future I envision involves an assistant that possesses in-depth knowledge of my life, functioning like a personal PA, but solely tethered to my devices and with my permission. That’s the LLM Siri Apple is currently developing, and even though I yearn for these capabilities immediately, I truly believe the wait will be rewarding.
An Extended Wait Ahead
Originally, I anticipated a launch for conversational Siri in early 2026, which already felt like a significant delay. However, recent updates from Bloomberg indicate that the rollout has been postponed to 2027—or potentially even later!
According to Gurman, employees within Apple’s AI division now believe that a more conversational version of Siri will not reach consumers until the launch of iOS 20 ‘at best.’
Heightened Expectations Ahead
Two universally applicable truths come into play concerning prolonged waits: firstly, the waiting period is painful and seems endless; secondly, once it concludes, the excitement of acquisition quickly overshadows the discomfort of the delay.
Fast-forwarding to the moment conversational Siri finally arrives on our devices, and assuming it genuinely meets all our expectations, we may find ourselves reminiscing about how long the journey took, but all our grievances will likely fade away.
However, the prolonged delay significantly amplifies the expectations for what Siri must achieve in 2027. Consider the advancements that ChatGPT, Claude, Gemini, Llama, and DeepSeek will have made by then!
Additionally, think about Amazon’s upgraded conversational Alexa, which will also possess two more years of development and an amassed wealth of user request data.
Siri will no longer be compared to today’s chatbots; rather, it will be evaluated against those available in two years’ time. The benchmark will be incredibly high, and Apple must truly meet it.
Image: Michael Bower/DMN
