In many ways, the relationship between AI and the IoT has many codependent features. However, it could also be easily argued that these features combine to make a sum more powerful than the whole – making AI and the IoT something of a power couple.
Grappling with the huge amounts of data the IoT produces
It was estimated that in 2024 there were approximately 18.8 billion connected IoT devices worldwide. This number is expected to continue growing, with some estimates suggesting the number of connected IoT devices will reach 40 billion by 2030.
18.8 billion devices produce a lot of data.
What happens to this data? In order to make use of the data these devices generate and convert it into useful, actionable information, its owners must invest in significant data storage and analytics capabilities.
This is where AI can step in! AI can assist with processing and analysing the data. Using AI, data scientists can automate the work of cleaning and organising the data. They can also deploy AI to run analyses and formulate suggested outcomes and recommended actions. Further, AI-powered modelling can be applied to provide predictive analytics.
Feeding the AI models with the data they need
In return, the data from our connected IoT devices is needed by AI.
Every AI model requires training data. The data generated by IoT devices can be very useful here. Of course, there are some caveats too: you need to ensure you have the right data to answer the question you are asking. Your data needs to be relevant to the use cases in which the AI models will be applied. And, of course, it needs to be accurate and complete.
Understanding these concepts falls into the realm of data scientists. It is they who feed the IoT data to the AI models in order to train them and test the modelling.
Even once successfully trained, AI cannot replace humans altogether. Every framework or set of guidelines for the responsible use of AI specify the need for a “human in the loop”. In other words, there still remains a need for human judgment; the ultimate decision making should rest with the human operators.
How does this beneficial marriage impact others?
Together, AI and IoT can automate many tasks. For example, in agriculture, IoT sensors can monitor soil conditions, while AI can determine the best times for watering and fertilizing crops, leading to higher yields.
In manufacturing and industrial environments, IoT devices monitor machinery in real time. Increasingly, AI algorithms are used to predict when maintenance is needed before a breakdown occurs. This prevents costly downtime and extends the life of equipment.
Even in our everyday lives, the partnership of AI and IoT can create more personalised and responsive user experiences. For example, AI can learn a user’s preferences for music, lighting and temperature and then IoT devices can be used to adjust settings accordingly to create the perfect ambiance.
Which bumps will AI and the IoT need to navigate through their relationship?
Given the staggering amounts of data that AI-IoT systems collect and process, there are data privacy and information governance concerns which must be addressed. As well as potentially including personally identifiable information – and therefore falling under strict data governance rules like UK GDPR – the data collected might open up other privacy issues. In some applications it could be used to track our movements, monitor our behaviours or even predict our future actions!
Bias is another key concern wherever AI systems are deployed. If the data on which the AI systems are trained is biased, the AI system will also be biased. Care needs to be taken when developing and testing AI models to identify possible sources of bias and engineer out their effects.
Plus, as with any IT system, AI and IoT devices can be subject to cyberattacks. If these systems or devices are hacked, they could be used to steal data, disrupt infrastructure or even cause physical harm. If AI is making autonomous decisions and controlling IoT devices which take real-world actions, the potential for disruption is significant. Owners must take considerable care to secure their devices, networks, data and integrations throughout the supply chain.
