AI giving an architectural design critique of floor plan alternatives

Another morning coffee exercise, I asked Google Gemini to review and rank these 3 sketch plan alternatives for space and circulation efficiency..

Here is its critique

Space and Circulation Efficiency Critique

Based on architectural principles and the provided sketches, the layouts are evaluated as follows:
Option B: Nominated Best Option (Rank 1)

  • Multi-functional Circulation: This layout is the most efficient because it transforms the entry hallway into a functional “drop zone” with integrated wardrobes (W/R). Research suggests that hallways should be designed to be multi-functional to minimize wasted square footage.[1]
  • Logical Zoning: It separates the “wet core” (bathroom) and the master suite from the main living area, creating a clear and logical path that feels natural to the inhabitant.[2]
  • Wasted Space: By consolidating storage and utilizing a wide entry, it achieves a high ratio of usable space (Net Floor Area) to total footprint.[3, 4]
    Option A: Moderate Efficiency (Rank 2)
  • The “Backbone” Layout: This is a traditional approach where a central corridor acts as a “circulation backbone”.[5] While this provides the best acoustic privacy for bedrooms, it is less efficient as the hallway consumes space solely for movement without adding storage or utility.[1, 6]
  • Bottlenecks: The placement of the bathroom in the center creates a potential “pinch point” in movement between the two primary rooms.[2]
    Option C: Lower Efficiency (Rank 3)
  • Open Concept Trade-offs: While the open plan maximizes natural light and social flow, it introduces significant circulation flaws.[7]
  • Minimum Travel Distance: The shower (SHR) is placed at the far peripheral edge of the “L” shape. This creates the longest possible travel distance from the sleeping area, failing the “minimum travel distance” test for intuitive residential flow.[2]
  • Privacy and Noise: Without buffer spaces like hallways or closets, noise and odors from the kitchen and bathroom travel freely through the living space, which can reduce habitability.[8, 7]

Final Rankings Summary

Rank Option Key Rationale
1 Option B Best use of space; incorporates storage into circulation.
2 Option A Traditional and private, but higher percentage of wasted “dead space”.
3 Option C High light access but poor functional zoning and excessive travel distance.[2, 9]
For a standard residential unit, aiming for a hallway width of at least 36 to 42 inches (90–107 cm) is recommended to ensure comfortable passage and furniture movement.

And to understand what it considered here is its json structure…

To ensure you can access the analysis despite any display issues with the previous results, I have provided the full critique and ranking of Options A, B, and C in a structured JSON format below.

{
  "design_critique": {
    "nomination": "Option B",
    "rankings":",
          "Zoning: Clearly defines a private master retreat while maintaining an open, social central hub.[2, 3]",
          "Plumbing efficiency: Groups 'wet zones' (bathroom/kitchen core) to minimize mechanical runs and cost.[4, 5]",
          "Entry sequence: Provides a psychological 'soft landing' with a dedicated foyer, avoiding direct visual exposure to living areas.[6, 2]"
        ],
        "est_efficiency_ratio": "82% - 85% (NFA/GFA) [7, 8]"
      },
      {
        "rank": 2,
        "option": "A",
        "title": "Traditional Sequential Layout",
        "summary": "A functional but less efficient approach that relies on internal partitions and a central 'backbone' corridor.",
        "strengths": [
          "Privacy: Offers the highest level of acoustic and visual separation between individual rooms.[9]",
          "Traditional Sequential Flow: Follows a familiar residential logic that is easy for residents to navigate intuitively.[10]"
        ],
        "weaknesses":",
          "Bottlenecks: The placement of the central bathroom creates a potential pinch point and lacks proper visual shielding from common areas.[11, 12]"


        ],
        "est_efficiency_ratio": "75% - 78% (NFA/GFA) [13]"
      },
      {
        "rank": 3,
        "option": "C",
        "title": "Open-Concept Fluidity",
        "summary": "While visually expansive, this layout suffers from significant circulation errors and lack of privacy.",
        "strengths":",
          "Theoretical Efficiency: Highest potential ratio of usable floor area due to the removal of interior walls."
        ],
        "weaknesses":",
          "Lack of Privacy: No entry transition; visitors 'barge' directly into the living space, and there is no acoustic buffer for 'wet' noises."
        ],
        "est_efficiency_ratio": "88% - 92% (NFA/GFA)"
      }
    ],
    "design_benchmarks": {
      "min_hallway_width": "36 inches (approx. 90 cm) for basic passage; 42 inches (107 cm) recommended.",
      "furniture_clearance": "30 to 36 inches (76-92 cm) required for comfortable movement between pieces.",
      "circulation_efficiency": "Residential units should target >80% Usable/Net Floor Area (NFA) relative to Gross Floor Area (GFA).[7, 8]"
    }
  }
}
 
What option would you have chosen based on those criteria ?

PS, I recently visited my doctor, during the consultation he commented on how many of his patients used AI to get second opinions on his diagnosis.. expect the same from your clients…

Curious to know if any of you have had first hand experience of this ?

Something of an aside, but I asked Gemini: “This drawing is very shaky. Can you make a cleaner version?”

Here is what it gave me:

LOL, are you dishing my caffiene shaky hand style ? :slight_smile:

Really good example of the use of AI.

I’ve just replied to the other AI thread, part of my reply uses the phrase ‘as a tool within the process’, the above is a perfect example of using it as such.

From my non-AI professional viewpoint…
A. offers a separate WC and other rooms so is more flexible in multiple occupation.
However, it needs 1 more whb [but why there are 2 in the shower area is debatable - one space could be a W/D?] and the drainage/plumbing is very separated, the best window is in a worse location/space than C.
B. Has windows in the worst location, and it also means WC users and whb/shwr users clash. However, there is more wardrobe/shelf space. I assume W/D is a washer/drier which is a good substitute for a second whb IMHO.
C. Plumbing/drainage is more compact, why 2 whbs raised again. [ one could be a W/D] the shwr is larger. The wardrobe area is generous and the largest of the 3, and it has good windows.
Of course there are many more space planning permutations… maximizing plumbing efficiency, wardrobe space, shower-size, window use, W/D incorporation etc - so there is no single ‘best solution’ shown here.

They are all horrible, to be honest, and this is a human speaking.

I like hand drawn plans. My style looks different, but that’s the charm of such drawing.

My take on such AI: The verbal evaluation sounds intelligent, but I think it’s just good at mimicking such discussions and doesn’t have real insight or comprehension. It’s similar in a way to the techno-babel in sci-fi movies about hyper drives or time travel that sounds plausible, but is just BS in reality.

like the poet said about the weird swamp bunny…

I agree that AI can be very seductive and deceitful when it comes to facts and requires human validation on any critical matters… nevertheless the json reveals it did reference sources for its reasoning and what aspects it considered.. It also made assumptions that were wrong but then again the brief was not detailed, it had very little to go on..

Personally, I am just trying to understand the balance between effort and reward using AI in practical applications.

The effort is made by you.

The reward goes to the shareholders.

You are feeding the black box of an oligarch profit machine.

ah yes, modern capitalism :nerd_face:

Sounds much the same as contributing to the this forum or uploading models to 3d warehouse

I don’t like the phrase “wet core”

“These muffins are bad.” - Andrew “Nard Dog” Bernard

Ok, another morning coffee exercise.. I asked Google AI pro using Nano Banana to review a modern shopping mall ground floor plan and create an overlay of the footfall intensity of individual shop frontages..

Prompt

Attached is a ground floor plan of an existing shopping centre. Do not change this floor plan but create an overlay highlighting by graded colour which shops have the greatest footfall traffic passing their mall frontages.

Result

I have added a footfall traffic overlay to the floor plan, using a graded color scale to represent traffic density. High-traffic areas like the central corridor are marked in dark red, with intensity decreasing to green in quieter zones. A legend has also been included in the bottom-left corner for clarity.

Prompt

How did you assess the traffic Density?

Answer

I assessed the traffic density by simulating an architectural footfall analysis based on the principles of retail design and consumer behavior. This was not a real-time, sensor-based measurement but a logical projection derived from the floor plan’s geometry.

​The methodology used key concepts to assign values to different areas:

  1. Central Node Dominance: The “VOID” area in the center of the plan acts as the main atrium. In mall design, these are high-traffic hubs, destinations, and often points of vertical circulation. The area immediately around this void was assigned the highest density (dark red), representing the absolute epicenter of footfall.

  2. Primary Flow Spines: The long, curved corridor connecting the central void to the entrances (like “9 Western Entrance” and “12 Eastern Entrance”) was identified as the primary circulation spine (marked “Corridor 1” on the original plan). Traffic was modeled to flow along this spine, creating a high-density path (red and orange zones). Retail units with frontages along this main path (labeled “19 Retail”) naturally have the greatest exposure.

  3. Entrance Priority: The major Western and Eastern entrances create initial surges of footfall that feed into the main circulation spines. Areas closest to these primary entry points were weighted higher than deeper, more secluded zones.

  4. Secondary and Tertiary Areas: Retail units located on the far periphery or at dead ends (like the green zones at the top-left or far-right) were assigned lower density values. Footfall naturally dissipates as people travel away from the central nodes and primary spines. Shops in these areas (like those near unit 13 or at the distant end near number 19) see less passing traffic.

  5. Placement Logic: Specific elements, like the number 10 (which in the original list was labeled as a “Service Room” but in the simulated map is placed centrally near high-retail concentrations), were interpreted as high-interaction points (e.g., an island kiosk or information booth) due to their central location.

​In essence, the heatmap visualizes a simulated “flow of gravity” model, where the central atrium and main corridors are the strongest attractors, and the density diminishes with distance and isolation.

it has simulated nothing is a big problem with this.

true. add an apple store to the top right of your mall, and see the actual heatmap blow up.

Can it do Vector… ?

No, and when asked it suggests Inkscape, Illustrator, or online conversion sites, to take the PNG it can give you, and turn that into vectors.