On Mobile World Congress 2019 in Barcelona, AI was not the most conspicuous technology being discussed, but its pervasive presence was undeniably felt and the influence of AI is expected to only grow in the future. AI has the potential to increase efficiency and utility of cellular capabilities, allowing for improved navigation, enhanced productivity, and advances in specialized applications. But AI also comes with a great deal of accountability and risk that can be difficult if not impossible to manage.
Qualcomm’s Snapdragon 8 Gen 2 Modem-RF System provides a great example of the potential of AI, as it reveals the ability to optimize 5G modem connectivity, camera capabilities, sound, and security performance on smartphones. By many accounts, Qualcomm also created the first on-device demonstration of Stable Diffusion using an Android device. Initially limited to large data centers, this dynamic text-to-image generative AI foundation model is capable of creating photorealistic images within tens of seconds.
Alternate network architectures are being designed with the help of AI and improved computing capabilities, but there are still several challenges that remain for the full implementation of AI. For example, AI model training is most often done in large data centers due to increased processing capacity, but when it comes to inferencing, some AI applications are better run on edge computers and devices for lower latency and enhanced security. Similarly, AI has an amazing ability to find obscure long-tail content online, like the chatbot ChatGPT, but AI applications can still be too biased, inaccurate and unresponsible to be used in situations like CV screening and recruitment.
Google search is not the only area that AI can be applied, but tremendous caution needs to be taken when implementing the technology, especially in situations that require highly-subjective decisions. Examples of this would include patent essentiality checks for standards such as 5G, determining royalty charges for standard-essential patent owners, or interpreting complex instructions and providing sophisticated responses in essays. AI can never be any better than the data it trained on, and the training data itself may not be sufficient or accurate.
In summary, while AI has wide potential applications, it is still in its infancy in many respects. AI can be used effectively in areas like radio planning, digital twin recalibration, and even making photorealistic images, but its use in scenarios like patent essentiality or CV screening may prove to be inefficient or even dangerous. Therefore, tremendous caution needs to be exercised when utilizing AI.
Qualcomm is a leading global semiconductor company that has made great contributions in the development of the cellular industry. Their flagship Snapdragon 8 Modem-RF Gen 2 System is a great example of the company’s leadership role in AI technology, as well as their commitment to optimizing 5G, camera capabilities, secure communication, and more.
On the other hand, OpenAI is a non-profit artificial intelligence research lab that was founded by Sam Altman and Elon Musk. They are dedicated to developing AI technologies that are beneficial for the world and ensure their responsible use. They are known for their research on GTP-3, and have cautioned the public about its potential misuse.
