What Apple's Failures Say About AI's Limitations — With M.G. Siegler

Episode Summary

Executive Summary: The episode argues that Apple’s muted WWDC and delayed Apple Intelligence rollout reflect both Apple’s execution problems and broader limits in current AI products. M.G. Siegler says Apple overpromised last year, underdelivered, and then pivoted this year to UI over AI because it lacked a compelling AI story. The discussion widens to question whether assistants can truly become reliable, context-aware products, given today’s chatbots, memory features, and legacy-device constraints.

Main Topics: Apple’s underwhelming WWDC and AI retreat (Priority: 5/5): Apple’s developer conference emphasized a new UI ('Liquid Glass') and peripheral features while largely downplaying Apple Intelligence. Siegler argues the company had little meaningful AI progress to show, making the event feel like a retreat from last year’s promises. The gap between AI demos and productization (Priority: 5/5): The conversation revisits Apple’s 2024 AI presentation and the recurring problem that impressive demos do not translate into usable products. Siegler says he was skeptical from the start about Apple’s ability to ship the features it previewed. Whether AI itself has product limits (Priority: 5/5): The hosts debate if Apple’s failures signal broader limitations of current AI. They note that chatbots work well, but smart assistants across Apple, Amazon, and Google have struggled to become reliable everyday products beyond narrow use cases. Legacy assistants as an albatross (Priority: 4/5): Siri, Alexa, and Google Assistant created expectations and product baggage that make it hard to retrofit modern AI into older platforms and devices. These companies are now trying to graft new AI onto architectures built for older assistant paradigms. Privacy, memory, and user trust (Priority: 4/5): The segment questions whether users actually want AI systems to ingest their emails, photos, and messages. The speakers suggest memory/context features may backfire if they feel intrusive or unreliable, despite their usefulness in theory. Apple’s strategic pressure and acquisition options (Priority: 4/5): Apple faces flat iPhone sales, pressure on services revenue, tariffs, and antitrust risks. The discussion suggests acquisitions like Perplexity or Anthropic could help Apple gain AI talent, product direction, and search capability faster than internal development alone. Liquid Glass and the state of Apple’s culture (Priority: 4/5): The UI redesign was seen as visually appealing by some, but early user reaction was mixed to negative, especially around readability. The episode closes with concerns about leaks, leadership, and whether Apple’s culture is too rigid for the AI era.

Key Arguments: Apple’s WWDC score was generous at 2/5 because the company had so little AI progress to show; absent a UI refresh, the event could have rated a zero. Apple deliberately shifted attention from AI to UI because AI remains an exposed weakness and the company could not deliver a strong developer story. The 2024 Apple Intelligence presentation was not as revolutionary as people later remembered; even then, many observers called it underwhelming. Current AI is genuinely useful in chatbots, but assistants that combine email, messages, maps, photos, and memory still feel unreliable and may not be the best product shape. Big tech’s legacy assistant products made upgrading harder, because the new models are being forced into old devices and old user expectations. Memory/context features may create more discomfort than value if AI surfaces old data that users no longer want to be guided by. Apple’s biggest problem is not just AI competition but corporate pressure: it needs a new growth engine while facing flat hardware growth and services risk. Apple may need acquisitions to change its AI talent, mentality, and execution speed; Perplexity is suggested as a plausible target, while Anthropic may be too expensive or structurally difficult. The likely future of AI may be multi-device and interface-agnostic, meaning the best experiences could emerge in forms not yet obvious to consumers. Apple’s leaks and muted event atmosphere suggest internal culture problems, not just a temporary product miss.

Data Points: WWDC rating: 2 out of 5 stars - Siegler’s overall assessment of Apple’s 2025 WWDC event Potential WWDC rating without errors: 0 out of 5 stars - Siegler says the event could have been a zero due to Apple’s mistakes and lack of AI progress Apple Intelligence talk time: roughly 3 minutes - Apple spent about as much time on Apple Intelligence as it did on the phone app Phone app discussion time: roughly 3 minutes - Used as a contrast to show Apple’s odd prioritization at WWDC Polling result last year ('meh'): 50.4% - Alex referenced a prior WWDC poll on Apple’s AI presentation Polling result this year ('meh'): 81% - Alex’s current poll about WWDC reaction, showing a 30-point increase in 'meh' Increase in 'meh' sentiment: 30 points - Year-over-year increase in audience meh reaction to Apple’s WWDC Google search deal value: $20 billion per year - Potential Apple services revenue risk if Google’s default search payments are lost OpenAI memory/context scale: 1-2 million token context windows - Discussed as a limit/aid for assistants handling large personal datasets Apple buybacks: $100 billion two years in a row - Used to argue Apple has the capital to pursue a major acquisition like Perplexity Perplexity valuation: $14-16 billion - Referenced as a potentially affordable acquisition target for Apple iPhone launch price: $700 - Siegler’s anecdote about initially doubting the iPhone’s value WWDC atmosphere change: One year later, Jobs Theater was mostly empty - Compared with the prior year’s more energized event environment

Pivotal Quotes: "I think they were able to sort of smartly navigate down that path, giving people things to talk about when they weren't going to be talking about, again, that elephant in the room." — M.G. Siegler: On Apple shifting emphasis from AI to UI at WWDC "AI is a new UI, right?" — M.G. Siegler: On why current AI products may require new interfaces rather than being bolted onto old ones "if I can't read notifications anymore, then I might as well stop talking to women altogether." — Nikita Beer (quoted by Alex): A reaction to Liquid Glass readability concerns and notification transparency

Implications: Apple may need acquisitions, cultural change, and a clearer AI product strategy to stay relevant. More broadly, the episode suggests AI’s next breakthrough may require new interfaces and devices—not just better chatbots or retrofitted assistants.