Why Apple’s Create ML matters to your enterprise

Apple’s introduction of Create ML will make it easier, faster, and cheaper for enterprise users to develop machine learning apps for external and internal deployment.

Apple, macOS, iOS, Xcode, Core ML, Create ML, Artificial intelligence, machine learning
Reuters / Elijah Nouvelage

Create ML is Apple’s attempt to commodify some of the challenging machine learning tasks developers must otherwise resolve alone. Apple has chosen to leverage its existing ML technologies, as found in Siri and Photos, for this.

What is Create ML?

Focused at present on vision and natural language data, developers can use Create ML with Swift to create machine learning models, models which are then trained to handle tasks such as understanding text, recognizing photos, or finding relationships between numbers.

It lets developers build machine learning models on their Macs that they can then deploy across Apple’s platforms using Swift.

Apple’s decision to commoditize its own machine learning tech means developers can build natural language and image classification models much faster than the task takes if built from scratch.

It also makes it possible to create these models without use of third-party AI training systems, such as IBM Watson or TensorFlow (though Create ML supports only very specific models).

How does Create ML benefit us?

Training machine intelligence is time consuming, so it is noteworthy that Apple claims Create ML will dramatically cut the time it takes to create models.

The company cited Memrise, which cut the time it took to train an image processing model from 24 hours to just 18 minutes. That’s because it builds on Apple’s existing, widely deployed ML models for images and speech, I imagine.

Apple also offers Core ML. The difference between the two is that while Create ML lets you quickly create AI models on Apple’s platform, Core ML lets you bring models you’ve made outside of Apple’s ecosystem (such as inside TensorFlow) aboard.

How to use Create ML (abridged)

Here’s an incredibly simplified attempt at a description of how to use Create ML to build a machine learning mode. Apple’s workflow for this has three essential parts: Data, Training, and Evaluation.


You collect data for the model you want to build — images of apples and oranges, for example. Divide this data roughly 80/20 between Training Data and Testing Data. Once you’ve gathered enough data, you create a new blank (Mac) template in Xcode.


This is where Apple has done something smart. In Xcode, developers just type three lines of code, drop their training and testing data into the code, and Apple’s system will begin analyzing it all.


You’ll see a percentage appear to let you know how accurate the ML code is. Once it’s accurate enough for your purposes, you simply save the file and place it inside the app you trained it for.

Simplicity is complex

What’s liberating about Create ML is that Apple has made the process of building AI models much more approachable (though expert users can still use complex algorithms to do so).

It has also made it possible to create these models in familiar Apple development environments, Xcode, Swift. You can also use Swift scripts to automate the creation and training of new models.

Another big advantage is ease of deployment. Once your ML model is working, you can integrate it into your apps by drag and dropping it into the application code.

Why Create ML benefits enterprise developers

There is a chronic shortage of highly skilled AI developers — these people are pretty much writing their own cheques. Yet, despite that shortage, there’s no sign of deceleration in terms of enterprises wanting to use AI technologies to benefit their business.

Apple’s introduction of Create ML makes AI development more accessible, albeit limited to vision and natural language implementations. (Google is also working in a similar direction with Google Cloud M and Swift for Tensorflow.)

This makes it possible for developers to more quickly build and deploy AI in their apps, which makes it feasible for enterprise users to experiment with machine learning within their own apps.

Enterprise developers who need to keep tight control of the data used to train their AI and want to avoid use of cloud services will also benefit, as does any enterprise focused on creating ML apps for their own increasingly iOS-based fleet.

Cutting development costs

While the implications for consumer-facing AI seem currently defined by shopping and the like, teams tasked with developing internal collaboration, customer support, or business management apps can now look to fast deployment of new machine learning models.

That the cost of development has fallen as a result of the move should also help nurture more experimental uses of ML technology across the board, potentially unleashing fresh innovation.

Naturally, as development moves forward, it’s possible enterprises will need to expand their teams with the addition of high-level AI experience, particularly as they seek to finesse their models for more robust performance in the real world.

I’m certain that large enterprises committed to major ML deployments will use virtualized cloud-based solutions to crunch through data to build their AI models — but Apple even supports those externally created models with CoreML.

Moravec’s Paradox

One thing Create ML doesn’t do is break Moravec’s Paradox that AI is better at high-level reasoning than it is at figuring out low-level sensorimotor skills.

Machine intelligence at present is usually just a combination of pattern matching alongside fragments of neural deep learning and a little automation.

However, knowledge is power, and Apple’s solution does mean we can all develop a better understanding of the potential of AI using tools we already own. So, if you’re brave enough to experiment with Xcode, you can build your own example apps following this simple guide here.

Google+? If you use social media and happen to be a Google+ user, why not join AppleHolic's Kool Aid Corner community and get involved with the conversation as we pursue the spirit of the New Model Apple?

Got a story? Please drop me a line via Twitter and let me know. I'd like it if you chose to follow me on Twitter so I can let you know about new articles I publish and reports I find.

Copyright © 2018 IDG Communications, Inc.

7 inconvenient truths about the hybrid work trend
Shop Tech Products at Amazon