In an attempt to start adopting the AI I keep talking about, I decided to turn the writing of a blog over to a machine.
Specifically, a company called Jasper AI.
Jasper calls itself “The #1 AI Content Platform for Businesses”. They also promise users can “Create content 10x faster with artificial intelligence.”
This was a massive undersell – it wrote me two blog posts in 30 secs. It normally takes me multiple hours to decide what to write about, procrastinate, research, procrastinate, write, procrastinate and finally post.
What was impressive about the AI-generated blogs?
The two best things about the AI blogs were the quality of the language and the structure.
It sounds as if it’s been written by a human. The language is completely natural and completely correct. There was not a single point in either blog where the grammar or vocabularly was slightly wrong. That’s certainly already an improvement on my writing!
Secondly, both blogs had a very clear and sensible structure. Quite simply an introduction, three main points, and a conclusion, neatly summarising the main points.
If I hadn’t known, I would not have had any inkling that the blogs were computer generated.
What was less impressive about the AI-generated blogs?
However, before getting too excited about the power of AI, there were some points that need work.
Overall the blogs felt like they had been written by a human but a human lacking in critical thought. They feel like something I would have written at age 14 when I had no interest in the topic and hadn’t done my research.
What made me think this?
Lack of Real-World Context
Firstly the blogs were written as if GANs (a type of AI algorithm) were being used every day to much success in the real estate industry.
Whilst this may be possible in the future, it is not the case today.
The AI blogs were obviously missing real-life experience. For example, they wrote things like:
Q//With GANs, developers can test out different designs for a property before any construction takes place. This can save time and money by avoiding costly mistakes that might be made during the construction process.//
(Jasper AI 2022)
This sentence implies that property developers decide to build a building, start building and see what happens along the way.
We all know this is not true.
Developers go through hundreds of iterations of plans in order to avoid costly mistakes. GANs will not change that process. However, GANs may make some parts of the iterative process happen more quickly and may come up with a wider range of options.
The AI also displayed its lack of real world experience when it wrote that a virtual tour created by a GAN would be “indistinguishable” from a real-life tour.
This may be possible when the technology for virtual reality has significantly improved. However today, a virtual tour created by a GAN (or any virtual tour) is generally a poorer experience and therefore highly distinguishable from real life.
Lack of Critique
The AI did not offer any critique.
The AI blog did not explain the problems that might be encountered using GANs or why they cannot perfectly design houses at the moment. It simply wrote as if GANs are being used effectively all throughout the real estate industry.
Lack of Nuance
The AI blogs almost could have interchanged the word “GAN” with “technology”.
The ideas it presented were not always specific to GANs. Mostly they wrote about how technology could be used to help the real estate industry, without explicitly explaining why GANs could do this better or worse than other methods we have today.
If I had just generated the blog and posted it without reviewing it, I would have been spreading poorly researched and in some cases, simply incorrect information.
Using AI to Support, Not Replace
One of the things I want to believe most about AI is that we can use it to support our human endeavors. It can do the basic boring stuff we don’t like doing and put us in a position to spend more time doing the things where we add unique value.
AI should augment humans, not replace humans.
This is what AI did for me here.
In the blog about using GANs in designing buildings (here), I used the base structure produced by Jasper AI. I added nuance and further information. Below are the original AI-generated posts so you can see how well-structured and well-written they are.
Teaching the Humans
I should caveat that this was my very first time using AI to write a blog. I may not have given the computer the right instructions. I simply asked the AI
“how can we use GANs in real estate”
Perhaps to get the nuanced & critical answer I wanted, I should have asked the AI
“provide a nuanced critique on why we can and cannot use GANs in real estate”
I just assumed that critique and nuance would have been included in any answer.
To use AI to support us in the best way, we will need to work out how to “speak” to the AI. This might become a “skill” that we put on our CVs just like “management experience” or “competence with Microsoft Office”. A good analogy I read was from Bill Cusick of Stability AI who was quoted in Dezeen:
Q// Software is like chess, in that it takes a short amount of time to learn but decades to master.//
(Cusick in this article.)
It took me literally seconds to ask Jasper AI to write me a blog. Perhaps it will take me years to learn how to ask it for a blog that I won’t need to edit.
Teaching the AI
The AI-generated article demonstrates how well computers have learned to replicate human-level speaking. This shouldn’t be that surprising as they have learned from humans, just like children learn from adults.
If we follow this trajectory, then it is essential for humans to create high-quality, thoughtful and nuanced content on the internet. If AI only sees people on Twitter attacking one another with crude words, AI will produce equally crude content. If AI sees us interacting respectfully and intelligently, then it will act in the same way.
This is one of Mo Gawdat’s central ideas in his book Scary Smart (great read – here’s the link).
Gawdat points out that:
Q//Every year we create more content on the internet than all of the knowledge created since the dawn of humanity./
(Gawdat 2021)
This means that we all have the opportunity and the responsibility to influence AI for the good. It’s not just down to the engineers. They create the algorithms that teach AI how to learn, but it’s from the content on the internet that AI will find what to learn.
If a bot now reads my blog below on real estate & GANs, which has more nuance and detail than the blog Jasper AI produced, perhaps next time someone asks Jasper AI about real estate & GANs it will have scanned the internet, absorbed the content of this blog and reproduce one with more nuance.
Below are the two blogs that Jasper AI produced in about 30 seconds. Here is the link to the blog that I wrote, building upon the core structure that Jasper AI produced for me, below.
AI-produced blogs start here:
AI-produced Blog 1
Blog Title: How GANs Are Changing the Real Estate Industry
Blog Introduction: You may have heard of GANs, or generative adversarial networks. These are artificial intelligence algorithms that can generate new data based on existing data.
For example, a GAN could be trained on a dataset of images of houses, and then generate new images of houses that look realistic but don’t exist in the real world.
GANs are changing the real estate industry because they can be used to generate realistic 3D models of houses and other buildings.
This is helpful for two main reasons.
First, it allows potential buyers to virtually tour a property before it is even built.
Second, it allows developers to test out different designs for a property before any construction takes place.
Blog Body:
The Benefits of Using GANs in Real Estate
There are many benefits to using GANs in the real estate industry. Here are just a few:
- Save Time & Money: With GANS, developers can test out different designs for a property before any construction takes place. This can save time and money by avoiding costly mistakes that might be made during the construction process.
- Increased Efficiency: By generating realistic 3D models of properties, GANs can help speed up the sales process since potential buyers will have a better idea of what they are interested in before even stepping foot into a property.
- Improved Customer Experience: Allowing potential buyers to virtually tour a property before it is even built can improve the customer experience by giving them a first look at a property without having to schedule an appointment or wait for the construction to be completed.
Conclusion:
Overall, GANs offer a lot of potential for the real estate industry. They can be used to save time and money, increase efficiency, and improve the customer experience. If you’re in the market for a new home, keep an eye out for properties that have been created with the help of GANs!
AI-produced Blog 2
Blog Title: How GANs Could Change the Face of Real Estate
Blog Introduction: At their core, GANs are a type of artificial intelligence that can learn to generate new data based on a set of training data. For example, a GAN trained on images of houses could learn to generate new, realistic images of houses. This has all sorts of potential applications in the world of real estate. Let’s explore a few of them.
Blog Body:
- Virtual House Tours: GANs could be used to create virtual house tours that are indistinguishable from real-life tours. This would be especially useful for properties that are difficult to access or far away from potential buyers.
- Generate New Designs: If you’re not satisfied with the current crop of home designs on the market, GANs could be used to generate entirely new ones. This would be especially useful for custom home builders who want to offer their clients a unique product.
- Create 3D models: GANs can also be used to create three-dimensional models of houses. This would be helpful for architects and developers who want to get a better sense of how a proposed development will look and feel.
Conclusion:
As you can see, there are all sorts of potential applications for GANs in the world of real estate. As this technology continues to develop, we can only imagine what other ways it will be used to change the face of this industry.
People | Spaces | Places 
1 Comment