Experiencing Artificial Intelligence in a New Way through On-Device AI

During the recent Mobile World Congress, there was a lot of talk about AI. The same buzz was also present at CES a couple of months ago: The primary focus of the top consumer tech expo was on how AI was integrated into every product. However, the excessive excitement can make it challenging to discern what to anticipate, what to be wary of, and what to disregard as a passing trend.

Chief Analyst Ben Wood of CCS Insight described the MWC atmosphere as “omnipresent … but also overwhelming.” He felt that the levels of AI at the event were becoming so high that they risked causing fatigue among attendees.

Despite this, there were some positive aspects. Wood stated, “The most remarkable demonstrations came from companies illustrating the advantages that AI could bring rather than just touting a product as AI-ready.”

Just a year ago, at MWC, the popular generative AI tool ChatGPT was in its infancy, and on-device AI was still just a concept for the tech companies in attendance. Fast forward to this year, and on-device AI is now a reality, with attendees like myself getting hands-on experience at the event.

I had the opportunity to try out several demos featuring AI on devices, and the most impressive ones truly brought artificial intelligence to life in ways previously unseen. Many products we are already familiar with, such as smartphones and cars, are being rejuvenated by AI, offering unique uses to distinguish themselves from competitors. Additionally, new AI-focused wearables and robots are emerging, potentially displacing the technology we currently know and love.

Above all, it was evident that on-device AI is not just a futuristic technology. It is available here and now and could influence your decision on your next tech purchase.

The era of AI phones has dawned

One key takeaway from MWC was that while all tech firms now have a plethora of AI tools at their disposal, they are implementing them in diverse ways.

For instance, Samsung has introduced Gauss, its proprietary large language model, to focus on translation on the Galaxy S24, while Honor integrates AI for eye-tracking on its newly revealed Magic 6 Pro. Oppo and Xiaomi have on-device generative AI that enhances phone cameras and photo editing capabilities.

This demonstrates that we are entering a phase of experimentation as companies explore the potential of AI and how it can enhance user experiences with their products.

Samsung’s Y.J. Kim, an executive vice president at the company, and head of its language AI team, emphasized the importance of delivering tangible benefits to users through AI tools on the Galaxy S24. “We must ensure that customers see real value in their day-to-day usage of our products and technologies,” he explained.

Moreover, there is collaboration among phone makers in utilizing AI tools due to shared partnerships. Google, as the provider of Android, is actively experimenting with AI features that will be available on devices from Samsung, Xiaomi, Oppo, Honor, and others.

At MWC, Google showcased its latest AI features, such as Circle to Search, a visual search function that lets users draw a circle around an item on the screen to search for it.

Demo of Circle to Search on three large phone screens Demo of Circle to Search on three large phone screens

Google’s Circle to Search feature was a notable highlight at their MWC booth.

Katie Collins/CNET

Another common partner among phone manufacturers is chipmaker Qualcomm, whose chips powered a wide range of devices at MWC. Their Snapdragon 8 Gen 3 chip, unveiled in late 2023, is found in many phones utilizing on-device generative AI.

Qualcomm’s Ziad Asghar, in charge of the company’s AI product roadmap, mentioned that just a year ago, they showcased a rudimentary demo of generative AI on a phone. Today, phones equipped with this technology are available for purchase. Asghar emphasized, “We provide the tools, and each of our partners can customize unique experiences that they believe are essential for their customers.”

At MWC, Qualcomm introduced its AI Hub, offering developers access to a variety of generative AI models that can be integrated into their products. This selection will expand, enabling companies using Qualcomm chips to incorporate diverse AI features.

The Galaxy S24 Ultra featuring the new chat translation feature The Galaxy S24 Ultra featuring the new chat translation feature

The Galaxy S24 Ultra showcasing the new chat translation feature.

Lisa Eadicicco/CNET

In addition to determining which AI features to develop, one of the key challenges for phone manufacturers will be making AI accessible on their more affordable devices. AI is currently primarily reserved for high-end phones like the Galaxy S24, but this will eventually change. The technology will trickle down to a broader range of devices in the future.

According to Asghar, there will inevitably be variations in the capabilities and performance between the most expensive and the budget-friendly devices, similar to what we see with camera technology on phones today.

AI is revolutionizing our interaction with technology

While AI enhancements to phones are promising, we are already witnessing AI being leveraged in ways that could fundamentally change how we engage with our devices and potentially alter our device preferences.

Qualcomm’s technology, aside from enabling companies to integrate AI into existing device lineups, is also powering concept phones like the T Phone created by Deutsche Telekom and Brain.AI. These partners utilized Qualcomm’s chipset to reimagine the phone interface, creating a seamless, app-less experience that adapts to user needs and tasks in real time.

Phone screen displaying flight options Phone screen displaying flight options

This real-time interface was generated based on a request to find flights.

Andrew Lanxon/CNET

In a demo I witnessed at MWC, AI showcased the potential to eliminate the need for constant app-switching when making plans or completing tasks. Asghar commented, “It fundamentally transforms the way we interact with devices, making it more intuitive.”

Furthermore, he expressed his desire to see this concept applied to mixed reality glasses, enabling new inputs through gestures, voice commands, and visual cues, reducing reliance on traditional screen interactions. Asghar remarked, “Technology is more engaging when it works seamlessly in the background, solving problems without being intrusive.”

His sentiments echoed Google DeepMind CEO Demis Hassabis’ question during the MWC keynote about the future of phones as the perfect form factor. Hassabis pondered, “In five years or more, will the phone remain the ideal device?” suggesting there are exciting innovations on the horizon.

As seen at CES with the Rabbit R1 and at MWC with the Humane AI Pin, these innovations are already taking shape. In my interaction with the AI Pin, a wearable device devoid of a screen that relies on voice and touch inputs, it was evident that AI is fostering a space for experimentation, raising questions about the next dominant piece of technology in our lives.

Watch this: Humane AI Pin Hands-On: Tiny Wearable Phonelet Beams Light Like R2-D2

AI is also opening up avenues for existing technology that hasn’t fully resonated with consumers beyond niche use cases.

Many individuals have engaged with generative AI chatbots like ChatGPT and are becoming accustomed to AI assistants. One standout company, Integrit from South Korea, presented a robot at the event showcasing how these services might be used in public settings such as hotels or stores. Their AI and robotics platform, Stella AI, features a large display on a robot arm that can swivel to face users directly.

Unlike previous service robots like Pepper, Stella is integrated with advanced AI models like OpenAI’s GPT-4 and Meta’s Llama, enabling sophisticated multilingual conversations. Rather than sporting a humanoid robot face like Pepper, Stella employs generative AI to project a lifelike human image on its display. The preference for human-like interaction, even if virtual, over humanoid robots remains uncertain at this early stage.

Leave a Reply

Your email address will not be published. Required fields are marked *