What is Generative Design? Key Insights from the first ADDD Community Webinar

The first ADDD (Automated Data-Driven Design) Community webinar brought together experts to explore Generative Design and its impact on architecture. Led by ADDD founder Allister Lewis, the discussion featured insights from Arvinder Singh, an architectural designer developing AI tools, and Leo Pickford, an IT manager with extensive AEC industry experience.

Understanding Generative Design vs. Generative AI

A key distinction emerged between Generative Design and Generative AI. As Arvinder explained, Generative Design is rule-based and deterministic - the same inputs will produce the same outputs. It allows users to define specific parameters and optimise designs based on those constraints. In contrast, Generative AI is less predictable, with the same prompt potentially yielding different results each time. The training data is also predetermined and inaccessible to end users, offering less direct control. Generative Design software generally to date has mainly been based on rule based algorithms and coding and therefore do not employ AI. However, more tools are being developed that use the accessibility of AI to augment their offering.

Benefits and Applications

Generative Design tools enable architects and developers to:

• Quickly explore multiple design options

• Optimise for specific factors like space utilisation, sustainability, and cost

• Generate iterations based on defined parameters and requirements

• Test different scenarios in early project stages

The technology is particularly valuable for early-stage feasibility studies and competitions, where rapid iteration can lead to better decision-making. While some architects argue they can produce designs faster manually, generative tools excel at creating numerous variations quickly, especially for preliminary planning.

Implementation Challenges

Several key challenges were identified for practices looking to adopt Generative Design:

Cost and Return on Investment (ROI): Many tools use subscription-based pricing models, which can become expensive as team size grows. Practices need to carefully evaluate the return on investment before committing.

• Technical Infrastructure: As Leo highlighted, IT teams must consider hardware requirements, deployment processes, and integration with existing systems. Web-based tools can simplify implementation compared to installed software.

• Interoperability: Data exchange between different platforms remains problematic. While some tools offer multiple export formats, others are more closed systems. Solutions like Speckle are emerging to help bridge these gaps.

• Maturity Levels: The market includes both established players with mature offerings and newer entrants still developing their capabilities. Understanding a tool's maturity level is crucial for risk assessment.

Finding the Right Solution

With dozens of Generative Design tools available, practices need guidance in selecting appropriate solutions. Key considerations include:

• Specific use cases (residential, commercial, parking structures, etc.)

• Budget constraints, especially for smaller practices

• Required features and functionality

• Export/import capabilities

• Terms of use and IP ownership

• Level of vendor maturity and support

The ADDD community is developing resources to help practices navigate these choices, including a comprehensive software review report launching in March 2023. SIgn up at www.addd.io/contact to receive updates on this.

Future Outlook

Several trends are shaping the future of Generative Design in architecture:

• Standardisation: There's growing recognition of the need for standardised file formats and data exchange protocols to improve interoperability.

• Sustainability Focus: Tools increasingly incorporate sustainability metrics and optimisation, supporting better-performing buildings from early design stages.

• Accessibility: Web-based solutions are making the technology more accessible to smaller practices that may not have extensive IT resources.

• Data Ownership: Questions around intellectual property rights and data ownership will likely lead to evolving legal frameworks and business models.

Practical Recommendations

For practices considering Generative Design adoption:

1. Start with clear use cases aligned to business needs

2. Engage IT teams early in the evaluation process

3. Consider total cost of ownership, including training and support

4. Test available demos and trials where possible

5. Review data ownership and export capabilities carefully

6. Plan for integration with existing workflows

Summary

The webinar highlighted both the transformative potential of Generative Design and the practical challenges of implementation. As the technology matures and becomes more accessible, practices of all sizes will need to carefully evaluate how these tools can enhance their workflows while managing associated risks and costs.

The ADDD Community will continue exploring these themes in future sessions, including hands-on demonstrations of specific tools. For architects and designers interested in learning more, joining the community provides access to ongoing discussions and resources around the evolution of automated design technologies.

See the full recording here: https://youtube.com/live/TpPDg9SaM88