Mads Jensen argues that generative AI is changing how things are made in three ways, by automating requirements synthesis, design and production, and predicts that every industry that makes something will be nearly unrecognisable within a decade.
- The post models making anything, from toasters to buildings to software, as three steps: synthesise requirements, design, and make the thing.
- AI can sift through large volumes of data, such as thousands of user tickets, to spot trends and help designers and product managers organise requirements.
- Generative AI can turn natural-language requirements into screen designs and then into back-end architecture, accelerating the design process.
- Automating the translation of designs into production processes, such as shop drawings or machine parameters, is the third and least developed step.
- Shorter development cycles cut costs and lead to better outcomes because designers can explore a wider design space at lower cost.
Here is a simple model for how we make “things” (toasters, iPhones, buildings, software):
- Collect and Synthesize requirements
- Create Conceptual & Detailed Design
- Make the “thing”
Generative AI is changing “the making of” in three ways
- (Semi) automating the synthesis of requirements
- Automating the translation of requirements into designs
- Automating the actual production
Let’s unpack
1. Automating the synthesis of requirements
Today the lead designer/product manager synthesises data into specific requirements. Getting to a good design often requires vast amounts of input data. And the best designers do a great job at combining new research with prior knowledge.
But AI algorithms can make a big impact here.
For instance, AI can sift through reams of data (e.g. thousands of user tickets) to spot trends. This can help organise requirements and empower designers and PMs. E.g. – leaving them to focus on higher-level prioritisation.
2. Automating the translation of requirements into designs
Historically, humans have translated requirements into designs/products/code.
With generative AI, computers can create new designs with natural language as input. And they can also create output based on other inputs – e.g. a visual design. For instance, a new workflow could be like this:
- the designer describes requirements in English
- the AI translates the requirements into screen designs
- the AI then translates the screens into the necessary back-end architecture
A generative AI model can theoretically use anything as input. This can accelerate the design process to whole new levels.
3. Automating the actual production
Today, human specialists translate designs into a production process. Shop drawings for architects. Machine parameters for factories. Some things are automated, but there is so much more that can be done.
And once we automate this third step, we will have transformed the way we make things.
Widget, meet your “AI Maker”.
Generative AI will serve to reduce the product development lifecycle. That reduces costs. But it will also lead to better outcomes. Why? Because designers will be able to explore a much wider design space at a lower cost.
We are already seeing a lot of automation across the manufacturing life-cycle. But on a grand scale, we are only getting started.
My prediction is as follows: every industry that makes something will be nearly unrecognisable within a decade. Be it buildings, products or software.
As always, this presents an opportunity. But it is also a big shift to existing industry and socio-economic structures.
Let’s embrace this wave of opportunity. And let’s also prepare society to support the transition in a sensible way.
Frequently Asked Questions
What are the three ways generative AI changes how things are made?
According to the post: (semi) automating the synthesis of requirements, automating the translation of requirements into designs, and automating the actual production.
Why would generative AI lead to better products and not just cheaper ones?
Because it shortens the product development lifecycle and lowers the cost of exploring designs, so designers can explore a much wider design space.
What risks or responsibilities come with this shift?
Mads Jensen notes that alongside the opportunity, the shift disrupts existing industry and socio-economic structures, so society needs to prepare to support the transition sensibly.