Without last week’s Silicon Valley Bank (SVB) collapse, AI and chatbots would dominate tech conversations. Microsoft-backed OpenAI released GPT-4, a new language model, in recent days. Anthropic’s Claude chatbot competed. Gmail and Docs are getting AI from Google. Chatbot-enabled Bing search has garnered attention. Missing name? Apple.
The Cupertino-based company held an internal AI and large language model event last month. The New York Times reported that Siri and other teams regularly test “language-generating concepts.”
Siri has long been criticized for misinterpreting queries, including mine. Siri and other assistants like Alexa and Google Assistant have trouble understanding the accents and phonetics of people from different countries who speak the same language.
ChatGPT’s and text-based search’s popularity makes AI model interaction easier. Enabling accessibility settings is the only way to use Siri.
In an interview with the NYT, former Siri engineer John Burke said that Apple’s Assistant has had a slow evolution due to “clunky code” that made even basic feature updates difficult. He also mentioned Siri’s massive word database. Engineers had to rebuild the database to add features or phrases, which took six weeks.
Apple’s language model strategy was unclear from the NYT report. Like Google and Microsoft, Tim Cook’s company wouldn’t limit itself to a Siri-powered chatbot. Apple has long supported artists and creators, so it would naturally apply language model advances to those areas.
Despite their initial invisibility, the company has been using AI-powered features for a while. Better keyboard suggestions, processing in photography, mask unlocks with Face ID, separation of objects from the background across the system, handwashing and crash detection on Apple Watch, and most recently karaoke on Apple Music None may be chatbots.
Apple rarely discusses its AI efforts. In January, the company began offering authors AI-powered narration to turn their books into audiobooks. This suggests Apple is considering generative AI applications. I expect to hear more about the company’s efforts in these areas at WWDC in a few months.