IoT Analytics report explores generative AI’s impact on IoT app development
IoT Analytics recently released a Generative AI Trend Report to understand the effects of generative AI on IoT devices and to learn more about specific use cases and IoT technology in general. The report looks into the competitive landscape for this technology and provided nine use cases of generative AI as it relates to IoT, three of which are highlighted here.
The report comes as ChatGPT, a large language model based on the GPT-3.5 architecture, is making waves in the world of artificial intelligence and machine learning. While its most commonly known use is for generating human-like text, ChatGPT is also being used for various other applications, including IoT code development.
Use Case #1: Code generation for IoT
The report says that large language models make code generation efficient for Internet of Things (IoT) applications possible. Generative AI is already integrated into many IDEs and can be a helpful tool for software developers. However, it is unlikely to replace humans shortly.
For example, GitHub Copilot is a cloud-based AI tool powered by OpenAI’s Codex model that provides real-time software code and functions suggestions. It can be integrated into various IDEs, such as VS Code or Pycharm and used to create IoT applications and devices. It works by analyzing the context of the file and related files to offer relevant suggestions.
Use Case #2: Robot control
According to the report, Generative AI can control autonomous (IoT) devices like robots by capturing motion data from animals or humans. This allows for complex movements of individual parts and for robots to make sense of their surroundings to reach higher horizon goals. Generative AI models help robots generate intermediate tasks without human intervention.
For example, Alphabet’s Deepmind has created a framework that allows robots to move and perform tasks with more natural characteristics. The report says this development brings numerous advantages, allowing robots to safely and efficiently traverse uneven terrain or handle delicate items without risk of damage or power depletion. The associated learning goals ensure consistent behavior in robotics applications.
Use Case #3: Social IoT devices
With the help of generative AI, connected devices can understand and respond to complex questions users ask. Users can also talk directly to their devices to customize settings and commands. Additionally, IoT devices can generate answers independently with the help of a large language model, which can be used to guide the robot if it is given unclear instructions.
For example, Amazon has developed the DialFRED framework, which allows robots to ask questions when they are uncertain. Through reinforcement learning, the questioner model is optimized to ask relevant questions at the right time, which can help in task completion. This framework includes an “oracle” that automatically answers the generated questions using correct data from the environment simulation.
Analysis
Market research firm Gartner has published predictions stating that by 2024, 80% of technology products and services will be built by people who are not technology professionals (such as software developers).
With the emergence of LLMs such as GPT 4, companies already targeting low-code/no-code environments are likely to be adding support for these models in order to support code generation within their platforms. As IoT Analytics points out, companies aren’t likely to move to an “autopilot” mode for code generation any time soon, but the incorporation of generative AI already looks to hold promise for developer productivity. By the same token, security teams will need to stay abreast of the increased output lest these systems regurgitate flawed code.
Article Topics
AI/ML | application development | AWS | ChatGPT | DevOps | Google | IoT | IoT Analytics | robotics
Comments