Hybrid AI and the Apple iPhone 15: A Glimpse into AI’s Future
Most AI implementations are cloud-based. Hybrid AI lets you split the AI processing load. We explore the advantages and use of this technology and what’s ahead.
- By Rob Enderle
- October 11, 2023
One of the recent topics being discussed by all the chip manufacturers and some of the hardware OEMs is hybrid AI. Every processor company is rushing AI components to market to provide for this hybrid concept, but because Apple is largely vertically integrated, it introduced its product and initial solution to market first.
In a few months, we’ll begin to see many more solutions, so let’s explore hybrid AI, why it is important, and how it could change how client devices work.
Hybrid AI
Currently, most AI implementations are cloud-based with some movement to put them on premises, but that is hybrid cloud, not hybrid AI. Hybrid AI is when you split the AI load between the central resource (be it cloud or company data center) and the device itself.
This is done for some of the same reasons hybrid cloud is done: to increase reliability and to lower latency. Although Apple’s initial use of this capability is kind of AI light, focused on Siri and enabling Siri to function when the phone is disconnected, the real value will emerge as we move generative AI to this hybrid architecture because we will increasingly rely on AI as a productivity tool, an advisory resource, and a critical operational element.
This means that the AI must be able to function if the network isn’t available or goes down. For instance, take a medical AI application supporting first responders in a catastrophe. Whether we are talking 9/11 when the communications networks (except for the old BlackBerry network) failed, or the huge Maui fire where the cellular networks went out of service, first responders dealing with burn and injury victims would be crippled if they were used to getting advanced AI support but were suddenly cut off from it.
In wars such as the current one in Ukraine, the AI must continue to function. Otherwise, a simple network outage could cause a critical effort to fail. We were given an example of this when Elon Musk allegedly cut off Starlink and crippled a major Ukrainian offensive operation using drones.
These problems would extend to autonomous cars, robots, and other machines that will increasingly need a functional AI to continue to operate, or where a lower level of latency than can be provided by a hosted AI is required.
Evolution of Hybrid AI
This initial Apple effort with Siri is binary, meaning it will work from the cloud or work from the phone as this technology advances. It will become more dynamic with loads shifting between the two extremes based on the needs of the device and the bandwidth and availability of the network. Future hybrid AI efforts will likely dynamically shift loads based on several parameters, which could include the loading of the platforms where the AI is running for better load balancing, better efficiency, and reduced cost.
This future hybrid AI solution should be far more robust, have reduced latency when needed, and still function if network access fails. The solutions will likely advance so that the user experience will increase more consistently across all the platforms (device, on premises, and cloud) to a point where the user may be unable to tell where it is being hosted.
However, we are far from that point today, with the only working example in production (that I know of) being the recently launched iPhone 15.
Final Thoughts
AI is undergoing massive change both in terms of what it can do and where it resides. These changes are only going to accelerate as we move away from the rudimentary consumer and digital assistant technology Apple just launched with the iPhone 15 and Siri to a full-blown AI solution that will be more robust, have much lower latency, and be far more useful despite network availability and access.
Don’t forget, at the same time that networks are advancing, we are deploying 5G now with 6G expected in just a few years. The future AI experience will have as much to do with the hybrid nature of the solution as it does with the software and hardware that defines it.
Apple gave us a glimpse of the future of AI, but what is yet to come will be incredible.
About the Author
Rob Enderle is the president and principal analyst at the Enderle Group, where he provides regional and global companies with guidance on how to create a credible dialogue with the market, target customer needs, create new business opportunities, anticipate technology changes, select vendors and products, and practice zero-dollar marketing. You can reach the author via email.