Apple is finally bringing AI to iPhone. Here’s what it’ll look like

Subscribe | Advertise

Good morning. It’s happening. Apple is bringing AI to iPhone. We break it all down for you below. 

In today’s edition: 

• 🌪️ AI for Good: Microsoft has a new weather prediction model 

• 💲 GenAI seed funding drops 76% from peak

• 🏢 Study: Young employees aren’t a (good) source of information about AI 

• 🍎 Apple is finally bringing AI to iPhone. Here’s what it’ll look like 

AI for Good: Aurora, Microsoft’s advanced weather prediction model 

Andrew Charlton-Perez, a professor of meteorology at the University of Reading, said in April that AI and machine learning are “transforming weather forecasting before our eyes.” 

Charlton-Perez led a study of the deadly Storm Ciarán that ravaged Europe last year; he found that the AI models tested were able to largely accurately predict the path of the storm up to 48 hours in advance. 

• Each of the models, however, underestimated the top speeds the storm’s winds reached. 

Microsoft’s Aurora: Microsoft’s AI for Science lab recently developed a new model — called Aurora — that was designed to overcome this specific challenge.

• The 1.3 billion parameter model was trained on “more than a million hours of diverse weather and climate simulations, which enables it to develop a comprehensive understanding of atmospheric dynamics.”

• Aurora’s comprehensive training, according to Microsoft researchers, makes it highly versatile; in addition to its more accurate storm tracking, the system can accurately forecast air pollution levels (up to five days in advance), which Microsoft called a “notoriously difficult task.”

Why it matters: One, this foundation model can enable a lot of different applications through fine-tuning, at a far less significant cost. 

And two, disaster preparedness impacts every industry and every person. The more we understand the specific destructive abilities of a given storm system, the better we can adapt, the better we can prepare and the more lives we can save. In the midst of our warming climate — which makes weather less predictable — this has never been so important. 

GenAI seed funding drops 76%

We’ve talked plenty about the idea that generative AI is growing in a hype-fueled bubble similar to the dot-com bubble of the late ‘90s. 

• Quickly, the notion here is that AI is super costly & unreliable (due to hallucinations and security concerns) which makes it tough for companies to generate a return on the enormous expense. Bubble, burst. 

This understanding is beginning to infect even the venture capital arena, which has been instrumental in hyping the boom AI. 

Key points: Pitchbook found that in the first quarter of 2024, U.S. VC deal value for pre-seed and seed-stage deals (for genAI) fell 76% from last year’s peak to $122.9 million. 

• “The latest quarter marks the lowest deal value and count since ChatGPT’s launch kicked off the generative AI frenzy in November 2022,” Pitchbook said. 

• That number was $517.7 million in the third quarter of 2023, the peak of the genAI hype cycle. It has fallen since then, landing at $372.4 million in the last quarter of 2023. 

Why it matters: Richard Dulude, co-founder of Underscore VC, told Pitchbook that the fundamental question is now around profitability. 

“I think in a lot of cases, people are going to be very disappointed,” he said. 

Together with Camunda

Best practices for incorporating AI in your automation strategy

AI offers numerous benefits -- from enhanced efficiency to improved decision-making. But it can also introduce plenty of risk if not executed properly.

Download Camunda's AI guide today to learn how best to thread that needle between risk (security, bias and ethics) and reward (wonderfully automated workflows).

Study: Young employees aren’t a (good) source of information about AI 

For years, senior employees in the workplace turned to their more junior counterparts for tutorials on new technology, according to a new study out of Harvard Business School. But the study found that when it comes to AI, this expectation of technological expertise shouldn’t really apply. 

Key points: The study is based on interviews with 78 junior consultants, who were interviewed following a field experiment with genAI. 

• The juniors were broadly excited to use genAI on the job, but anticipated that they’d have to teach their older colleagues how to use it. They specifically expected needing to alleviate managerial concerns about AI. 

• “The tactics that the juniors recommended to mitigate their seniors’ concerns ran counter to those recommended by experts in GenAI technology … and so revealed that the junior professionals might not be the best source of expertise in the effective use of this emerging technology,” the report reads.  

The big gap here revolves around the juniors’ “novice” recommendations for AI risk mitigation, which are “grounded in a lack of deep understanding of the emerging technology’s capabilities.”

Why it matters: As more businesses move to adopt generative AI, there are a number of risks to consider. This is another one, suggesting that companies (from management to older employees) should not rely on the “expertise” of younger employees, as they probably don’t have any. 

“Juniors may recommend novice AI risk mitigation tactics because juniors themselves may not be technical experts, and because when technology is nascent and exponentially changing, juniors may have had no formal training on how to use the technology,” the study reads.

💰AI Jobs Board:

• Research Scientist: Altera · United States · Menlo Park, CA · Full-time · (Apply here)

• AI Product Manager, GenAI: Scale AI · United States · Hybrid, NY · Full-time · (Apply here)

• AI Expert: JG Security, Inc. · United States · San Antonio, TX · Full-time · (Apply here)

 🕋 Events: *

Can’t Miss: Join Nvidia, OpenAI and Google at AIQCon

• Time is running out for you to snag tickets to attend Kolena’s AI Quality Conference (happening in San Francisco on June 25). In this jam-packed event, you’ll get the chance to engage with experts from the world’s biggest names in AI.

Learn, network and have fun doing it. TDV readers can purchase tickets here for 50% off using TDV50.

🌎 The Broad View:

• The rise and fall of Koo, India’s Twitter alternative (Rest of World).

• All the news from Microsoft’s Games showcase (The Verge).

• Toyota lost over $15 billion in market value last week after being caught falsifying tests (CNBC).

• Arm, Qualcomm legal battle could disrupt the launch of Microsoft’s AI-powered PCs (Reuters).

• Perplexity is scraping the work of journalists at Forbes, CNBC, Bloomberg and other pubs (Forbes).

*Indicates a sponsored link

Together with Tabnine

Generative AI that’s personalized to your teams

Software developers hate spending time on basic, repetitive tasks and on rote maintenance. So it's no surprise that they're flocking to AI code assistants to automate some of their work.

But general coding assistants aren’t nearly helpful enough. 

Tabnine, however, is the AI code assistant that’s personalized to you. 

• Tabnine connects to your codebase, learning from and adapting to you.

• Tabnine is context-aware, generating code that matches your best work.

• Tabnine’s models can be fine-tuned based on your code for maximum customization.

Join the more than 1 million developers who use Tabnine today to accelerate and simplify software development. Try it free for 90 days — or get a full year for the discounted price of $99.

Apple is finally bringing AI to iPhone. Here’s what it’ll look like 

For the past year-and-a-half, Apple has been the one mega-cap tech giant not obviously competing in the AI race that has so violently consumed its peers.

But on Monday, Cupertino finally said “AI,” and as The Information’s Jessica Lessin pointed out, differed immediately from just about everyone else by offering AI as a set of features, rather than a product in and of itself. 

Apple Intelligence, AI coming to IOS, iPadOS & MacOS Sequoia this fall: 

• Siri’s AI overhaul & ChatGPT integration

  • The gist of it is that Siri will be a lot better. Part of this involves an integration with ChatGPT. This integration will bring ChatGPT to the new operating systems for free, with Siri asking users if they want to use ChatGPT to answer queries. 

  • The other part of the overhaul involves an integration with each user’s personal device information. The example Apple gives is this: “A user can say, ‘Play that podcast that Jamie recommended,’ and Siri will locate and play the episode, without the user having to remember whether it was mentioned in a text or an email.”

• AI Writing tools will be a part of the ecosystem, across all apps, third-party or otherwise. 

• A ‘new standard for privacy’

  • Apple said “many” of the models behind Apple Intelligence run on-device, which is more secure. Cloud operations, according to Apple, will also be secure. 

The unanswered questions: 

The way that Siri will be able to understand personal user information seems somewhat similar to Microsoft Recall, which recently came under fire for a litany of severe security concerns. 

I asked Apple (and did not hear back) how this integration is technically being achieved — either through parsed screenshots or a sharing of user information with its AI models.

• I also asked about opt-ins or opt-outs … in other words, will users have the ability to turn these features off? 

And Axios reported that all the on-device AI models are proprietary; Apple’s LLM was partially built using public data from the web in addition to licensed content, though Apple provided no further details.

My View: First, I think many of these integrations (voice memo transcription, calling recording) just made the AI startup space that much more challenging.

And while many of the features here feel more usable than other AI products, IOS 18 could well underline the inescapability of genAI. And I’m not sure if everyone’s on board with that. 

I’ll follow up on this if and when Apple responds to my request for comment. 

• Refact.ai: AI coding assistant with code improvement tools. 

• Salt AI: A tool for AI workflow creation & scaling. 

• Deepgram: A platform for speech-to-text and text-to-speech conversion.

Have cool resources or tools to share? Submit a tool or reach us by replying to this email (or DM us on Twitter).

*Indicates a sponsored link

SPONSOR THIS NEWSLETTER

The Deep View is currently one of the world’s fastest-growing newsletters, adding thousands of AI enthusiasts a week to our incredible family of over 200,000! Our readers work at top companies like Apple, Meta, OpenAI, Google, Microsoft and many more.

If you want to share your company or product with fellow AI enthusiasts before we’re fully booked, reserve an ad slot here.

One last thing👇

That's a wrap for now! We hope you enjoyed today’s newsletter :)

We appreciate your continued support! We'll catch you in the next edition 👋

-Ian Krietzberg, Editor-in-Chief, The Deep View