Redefining the possible: how AI is transforming the world of product design
From robots that can play the piano and self-driving planes and boats, to a spice machine that offers a personalised cooking experience, and a toothbrush that promises to clean your teeth in 20 seconds – nearly every stand and product at the recent Consumer Electronics Show (CES) in Las Vegas had AI embedded in some way. The conference’s keynote speaker, Nvidia CEO Jensen Huang, told his 6,000-strong audience that it’s innovation that has happened at an “incredible pace”.
“It started with perception AI – understanding images, words and sounds. Then generative AI – creating text, images and sound,” he explained. Now we’re entering an era of “physical AI that can proceed, reason, plan and act.”
For hardware designers and engineers, AI is expanding the toolbox they have to play with. It’s creating new functionality and opportunities to enable greater personalisation. It’s optimising material choices and processing power. And it’s reducing the time, labour and cost associated with new product design. Devices that were once task specific are able to become more flexible, adapting to new demands.
Making next-generation hardware a reality
AI-assisted design stage
AI systems are being used to improve products at the start of the design process. Engineers are able to brainstorm ideas for the initial prompt, run thousands of simulations, and catch design errors and opportunities for optimisation earlier on. This leads to better accuracy, efficiency and lower project costs. In product research and design alone, McKinsey estimates GenAI could unlock $60bn in productivity gains. Its analysts have already seen a reduction of 70% in product development cycle times when AI tools are used.
Hardware optimisation
AI algorithms can optimise the allocation of resources, such as memory and processing units, and enable adaptations if and when product requirements evolve. AI can also identify issues early in the design process through the use of simulations and improve product reliability and security. NASA for example is using AI to create components that are significantly stronger, than previous designs, while saving two-thirds of the weight.
Software-hardware co-development
AI hardware and software are interdependent parts of the product that rely on each other to create an AI model that works. AI architecture such as graphics processing units (GPUs), compute processing units (CPUs) and Tensor Processing Unit (TPUs) requires high computational power, and the software must be designed to take advantage of the hardware it is integrated with. This is what makes products able to learn and adapt to individual user behaviour and preferences. Smart watches for example can track steps, blood pressure and sleep patterns, and smart thermostats are analysing building temperatures and improving energy efficiency.
Accelerated testing
AI is helping engineers replace many physical tests with virtual assessments that are faster and cost less, leading to more efficient and cost-effective hardware solutions. It’s also able to predict hardware failures and performance bottlenecks, enabling proactive maintenance and design improvements.
Not AI for AI’s Sake
While products with AI hardware can offer superior performance and customisation for certain applications, it’s still true that not every project requires AI capabilities. Regular computing uses algorithms to perform specific tasks within a set of pre-defined instructions, such as baking a cake with a recipe. In many cases, it is cost-effective and reliable enough for the task at hand.
When assessing whether AI is appropriate to include, there are a number of factors that need to be considered, including:
Data availability
AI algorithms need large, high-quality datasets for training and validation. Where this isn’t already present, the data collection work required can add significant time (and cost) to a project.
Privacy and security concerns
Customising hardware based on user data raises privacy issues, and design engineers need to ensure they are aware of the legislation on this issue. AI algorithms need to be ethically designed to prevent biases and unintended consequences. And cybersecurity needs to be baked into any solution.
Power consumption and heat
AI hardware can be power hungry and generate a lot of heat, which may not be suitable for every product design. Increasingly, engineers are being asked to scale down hardware without compromising performance, which creates added pressure when layering new levels of computational power into a product.
Supply constraints
Pressure on the global semiconductor hardware market may lead to shortages and supply issues in the near future. The US for example, has imposed restrictions on the export of AI chips to limit China’s access to AI computing power.
Integrating AI into existing workflows
AI can help at every stage of product design but integrating it into existing workflows can be complex and require significant changes to established processes. As is often the case when any process change is made, there’s also a risk it will cause manufacturing bottlenecks and delays with delivery.
Finding the right partner to guide you through
The world is on the verge of the next technological revolution. The potential of AI to reshape the world’s services and products is becoming a reality. It’s inspiring innovation and unlocking previously unimagined opportunities.
This is all within reach for every product business. But it requires an experienced team to guide them through. One that has the in-house expertise to manage the process end-to-end, and find the optimum solution for the end customer.
It’s only by working together that we can redefine the boundaries of what is possible.
Contact TG0 experts today to learn how our tech can lower energy requirements and chip usage in your AI-powered products!
