Durga Malladi, Qualcomm SVP and GM and "Godfather of 5G", discusses the future of AI in mobile tech
Sep 25, 2023
34:15
auto_awesome Snipd AI
Durga Malladi, Qualcomm SVP and GM and 'Godfather of 5G', discusses Qualcomm's history in wireless technology and AI roadmap, the future of LLMs running locally on phones, improving consumer trust in LLMs, and the transformative potential of generative AI in mobile tech.
Read more
AI Summary
AI Chapters
Episode notes
auto_awesome
Podcast summary created with Snipd AI
Quick takeaways
Qualcomm aims to bring AI processing closer to end-users by running inference locally on devices while training is done in the cloud.
Qualcomm acknowledges the limitations of large language models (LLMs) and aims to address issues such as bias, hallucinations, and limited training on non-text data to build trust in AI technology.
Deep dives
Increasing data transmission over networks
The podcast discusses the continuous increase in the amount of data being produced and transmitted over networks. This ever-growing data transmission has no end in sight. The focus is on determining what processing should be done in devices and what can be done in the cloud.
Qualcomm's role in technology advancements
The podcast introduces Durga Miladi, considered the godfather of 5G, who shares insights into Qualcomm's role in technological advancements. Qualcomm has been a major player in wireless technology, particularly in cellular networks, and has expanded their role to include processing power and AI. Their technology roadmap spans various business units and focuses on connectivity, AI, processors, and multimedia.
Shifting AI processing closer to end-users
The podcast explores the strategy of bringing AI processing closer to end-users. Initially, the focus is on running inference locally on devices while training is done in the cloud. This approach ensures improved user experiences, reduced latency, and better power consumption. The ultimate goal is to fine-tune models on devices based on personalized data, further enhancing the user experience.
Addressing limitations and the future of LLMs
The podcast acknowledges the limitations of large language models (LLMs), such as hallucinations, bias, and limited training on non-text data. Qualcomm emphasizes the importance of addressing these issues to build trust in AI technology. They aim to overcome these challenges gradually and ensure responsible AI deployment. The podcast predicts a transformative user experience and perceivable differentiation due to AI-enabled devices within the next decade.
Durga Mulladi is the SVP & GM for Technology Planning & Edge Solutions at Qualcomm Technologies, Inc., the iconic company best known for enabling cell phones via its CDMA technology and chipsets that were first demonstrated in 1985. Durga has been an integral part of Qualcomm's growth having spent nearly 26 years there in various technology leadership roles.
He holds 578 patents, is a senior member of IEEE, received his PhD in '98 from UCLA, and was awarded Qualcomm's IP excellence award. Durga's list of accolades and accomplishments goes on for days. We're all fortunate to learn from a wireless pioneer and true tech legend.
Listen and learn...
Durga's insights from more than 25 years pioneering wireless technology
An insider's view of Qualcomm's formula for success
How networks and chips enabled the birth of the smartphone
Qualcomm's AI roadmap
How soon we can expect LLMs running locally on phones
How AI takes advantage of the unique capabilities of 5G networks
How to figure out what transactions happen on the device vs. in the cloud
How LLM fine-tuning may soon happen on the edge of the network or on the device
What size LLMs can be run locally while managing power consumption