Welcome to this week’s edition of the Architecture Insights newsletter. Artificial intelligence for architects, landscape architects, and designers.

Edition #38:

• Blender AI Addons: Generative AI inside 3D model rendering software

• GPT4o Upgrades: Create tables and charts within ChatGPT

• Continued AI Safety Concerns: OpenAI employees signing letters against the company

• Image: A red walkway, green landscape.

Generative Design: Blender Addons

Blender, the well-known open-source 3D creation suite, has been integrating artificial intelligence through various addons. These AI tools are changing how architects, landscape architects, and designers work, helping them to be more efficient and creative.

Blender is a free and open-source software. Many add-ons are available for free, but some are paid.

Some of the best AI addons for Blender:

iCity:

iCity is a generative design tool that uses AI to create detailed urban environments. Think of it as a way to generate buildings, streets, and landscapes based on user-defined parameters.

iCity’s procedural generation capabilities streamline the workflow by eliminating the need for repetitive manual modelling. It has over 20 building variations and an extensive asset library.

The procedural road tool allows you to effortlessly craft roads tailored to your vision by adjusting width, lane count, and seamlessly integrating assets like sidewalks, intersections, lights, benches, and trees.

The imperfection system adds realism and character to the generated cityscapes by scattering garbage, leaves, puddles, and even adding water drainage for a truly lived-in feel.

Customizable building presets allow for a quick population of cityscapes with diverse architectural styles.

AI Render (with Stable Diffusion):

AI Render integrates Stable Diffusion directly into Blender. This addon allows users to generate high-quality images, animations, and objects from text prompts and scene data.

This free add-on enables you to enter text prompts and select from a variety of presets, including photography, portraits, watercolours, and design styles.

For architects and landscape architects, this means rapid visualization of design concepts and the ability to iterate quickly on ideas without extensive manual rendering.

Being able to create animations broadens its usefulness for all sorts of projects.

The learning process is a bit harder than other rendering software, but once you get used to it, you can do a lot with it.

ControlNet (with Stable Diffusion):

Blender-ControlNet integrates ControlNet by Stable Diffusion. This addon allows users to provide additional input, such as sketches or reference images.

This capability is particularly useful for creating concept art, landscape designs, and detailed architectural visualizations.

Tables & Graphs with GPT-4o

To create tables and graphs with your data using the new GPT-4o updates, you can follow these steps:

(Do not enter any sensitive information)

1. Access GPT-4o

As of May, GPT4o is free to all users.

2. Upload Your Data

You can upload datasets from sources like Google Drive or Microsoft OneDrive. This can include tables, PDFs, documents, or spreadsheets.

3. Generate Interactive Tables

Once your data is uploaded, you can ask GPT-4o to create an interactive table. This table will be customizable and expandable, allowing you to edit and ask follow-up questions by clicking on specific areas.

4. Create Charts

You can request GPT-4o to generate various types of charts, such as bar charts, line graphs, pie charts, and scatter plots. For example, you can upload a spreadsheet and ask GPT-4o to create a chart based on a specific statistic.

5. Customize and Download

After generating the charts, you can customize them by changing colours or elements. You can also download these charts.

Example Workflow

Here’s a step-by-step example of how you might use GPT-4o to create a bar chart from your data:

1. Upload your data (e.g., a CSV file) to ChatGPT.

2. Ask GPT-4o to create a bar chart. For instance, you might say, "Create a bar chart showing the distribution of building types in my dataset."

3. Once the chart is generated, you can customize it by asking GPT-4o to change colours or add labels.

4. Download the final version.

Follow-up Questions: Click on any area of the table or chart to ask follow-up questions and get more insights.

Example Prompt

"Please create a bar chart from this dataset showing the number of residential, commercial, and industrial buildings. Use different colors for each category and label the axes appropriately."

Continued Concerns On AI Safety

Last week, we discussed the safety concerns raised at OpenAI. This week, AI employees have intensified the issue by issuing an open letter calling for more attention to the safety of AI and AI employees.

In a joint letter, AI employees from OpenAI, Google, and Anthropic emphasize the need for caution as we advance towards artificial superintelligence. They highlight dangers such as amplifying societal biases, spreading weaponized misinformation, and the potential for an uncontrollable, Terminator-like AI scenario.

What They’re Asking For

• They want to eliminate agreements that prevent former employees from criticizing the AI companies they once worked for.

• A system for employees to anonymously raise safety concerns to higher-ups.

• They call for safeguards for whistleblowers who speak out about safety threats.

• AI companies prioritize transparency and open communication. They advocate for a culture where dissent is not suppressed and concerns about risks can be raised anonymously without fear of retribution.

In design terms, this would be similar to a leading multidisciplinary design firm having several employees quit and claim that their projects are unsustainable designs and do not prioritize serving humanity. Then members of other firms agree and sign a letter supporting their claims.

AI Image Of The Week