I was not an ambitious student at the beginning. At least in my first bachelor's degree at the University of Regensburg in business administration with a focus on real estate management at IREBS. At times, attending lectures was erratic and dialogues with my fellow students in the WiWi-Cafete more exciting. Project Development 1 with Prof. Bone-Winkel was one of the exceptions. In this course there was a guest lecture by the well-known architect Eike Becker. Almost 50% of us wanted to become developers designing architecturally valuable landmarks or revitalizing interesting areas. Some years later, I contacted Eike and asked him if we, two innovation enthusiasts, would like to talk. Thanks to LinkedIn. That's how we ended up in this upcoming conversation. According to the New Normal, it's not taking place in the WiWi cafeteria, but virtually. Fitting the virtual medium, artificial intelligence, the hype around ChatGPT and language models, respectively, and the impact on the real estate industry will be discussed.

Viktor: When did you have your first aha effect with AI?

Eike: During a colonoscopy, just before fainting, I hear the doctor in a good mood: " Ah, there's the AI reporting, there is something else I missed! Poof, it's gone!" That was a year ago. 

Viktor (with a laugh): Good introduction to the topic. My father, who is a doctor, will be happy about that. Did you also have such an experience at work?

Eike: I've been told that 95% of AI research is about making people healthier and extending their lives.
One of my coworkers recently wrote a particularly elegantly worded protocol. He was assisted by a text generation model (Natural Language Processing). To our amusement, he asked Chat GPT of OpenAI whether Eike Becker uses populism as a stylistic device in his columns. The answer consisted of eloquently worded, friendly catchphrases.
Meanwhile, we worked with AI when designing an interior and had it make suggestions based on room dimensions, budget, colors, styles and furniture types.
We've used the AI-based technology called computer vision to develop fairly systematic apartment floor plans and create sugary visualizations. Today, that only partially meets our needs, but it's evolving at an unprecedented rate. Using a text-to-image AI generator from Midjourney, DALL-E 2 or Stable Diffusion, words become images in minutes. This can be done playfully. In all fields, I expect changes in the way we work. AI can piece together something from a sea of known and make it more efficient at a rapid pace. Machine learning is a good term for it. All tasks, whether rather boring or rather creative, will soon be handled with the support of AI. But you are much closer to the topic, Viktor. In which direction do you think AI is developing?

Viktor: You should repeat the question today. As language models are improving substantially from version to version.
For example, in GPT-3, the question of whether lemurs that are provided with sufficient fruit like to work overtime in the office could still have led to the answer that lemurs like to work overtime in the office if they are provided with sufficient fruit only.
In the following model update, such a suggestive question would not have been so successful. Exemplarily, the answer would be: "Lemurs, as animals, do not work in the office."
In other words, the models are learning at breakneck speed, much faster than expected. Each benchmark is mastered faster and better. Now, for now, all applications that benefit from AI, that is, everything that processes data, will be equipped with AI, thanks to the OpenAI API and other large language models.

Eike: That's both amusing and amazing. Have you already worked with such models?

Viktor: For example, I coded a tool last week in one afternoon using ChatGPT, which I'm sure would have taken me three times as long just a few months ago. We are going to use the tool at Acquirepad to provide sellers of real estate with even better information about the companies that invite them to a transaction.

Eike: " It sounds good and positive, but do you see any risks?"

Viktor: The risk I see regarding this is that the AI solutions from providers such as Microsoft, Alphabet, Tencent, Alibaba, Baidu or Meta will fall under the rubric of "winner takes all" and the companies will build up an oligopoly that is almost impossible to overcome. This is part of the current AI development as well. On the other hand, there are also parsimonious language models that are open source and may be able to prevent this development. Remains exciting. If most people and companies start using an API from one provider or a few providers to equip their applications with AI, they feed training data and enter into a "lock-in". Also, you have to realize that just the previously mentioned companies have generated proprietary usage data from their existing applications that is not available to anyone else.

Eike: You expect that smaller companies, e.g. from the real estate industry, will not be able to compete.

Viktor: Startups and less data-oriented companies struggle to get ahead when this is happening. They don't have the data, the budgets, the intellectual property, and the talent in competitive numbers. But I see the option to develop new processes that are more data-efficient and similarly good or even better. I do not see regulatory intervention as useful at this time, but rather as inhibiting innovation. It would only be a last resort if oligopolies or monopolies were to form. This is my concern in these exciting times. Nevertheless, the productivity gains and the implications on research, consumption, health, jobs and the environment are unstoppable. What does that trigger in you?

Eike: Viktor, you fear, as most experts do, that the Googles, Microsofts, Tencents of this world will become so frantically smarter through AI that they will gain an unassailable advantage. Should you be right, I hope the US and Europe unbundle these monopolies. But things may turn out differently. Jürgen Schmidhuber, director at IDSIA, predicts that AI will soon be very cheap. So cheap that everyone will be able to afford this technology and then make it work for them on a variety of levels. I like that. Just the chance that these new brooms could help us sorcerer's apprentices make the world a better place gives me hope. In any case, the entire 25,000 laws, standards and rules, for example, that prescribe building in Germany are uploaded in no time at all. From that, again, a lot of useful applications can be derived.

Viktor: Good points. And I also believe that there will be many new business ideas. My statements have primarily referred to the basic models, which can then be used for new solutions.

Let's move on to the second part of the conversation.

Viktor: In the first part, you mentioned the topic of regulation for construction, where we can certainly still automate e.g. permits and planning. So what are your expectations, for example, of AI in architecture?

Eike: The demands on architecture and urban planning are becoming higher and higher, the tasks more and more complex, and the real estate industry is increasingly undermining these requirements. That's where we can make good use of AI support. Urban development, mobility, sustainability, density, social building, modular building, cradle-to-cradle, the organization of collaboration and approval processes could all get a big boost from AI and move forward for the benefit of all.

Viktor: I think so, too. You can pour all the individual requirements, regulations, even down to the municipal level, learnings from past projects, static restrictions and so on, into codes to develop recommendation systems. In the future, you describe an object and generate the design, right down to the detailed drawing. What else are you thinking about?

Eike: It will also be easier to develop high-rise cores, staircases, optimize the use of building materials, reduce steel and concrete, control technical systems more economically, and reduce technology in buildings. Through better analysis and optimization methods, there will be a significant reduction in mechanical technology. I call this Smartphone and Stupidhome. The real estate industry has not yet discovered the smartphone for itself and still thinks it's such a device for making phone calls.

Viktor: "Smartphone and Stupidhome" appeals to me. (laughs) So do you think that in the future people will control even more with the smartphone and less with buttons or built-in controls?

Eike: That's exactly what I mean.

Viktor: This requires an enduring communication standard, so that smartphones, which have much shorter life cycles with their operating systems, are also future-proof as smarthome components. Otherwise, there will only be more "planned obsolescence," as is unfortunately the case today with most technical components that cannot be repaired.

Eike: The industry today is trying to stuff buildings full of 20th century technology and blow as much air as possible through shafts and ducts using fossil fuels. That's outdated. The immense amount of energy used to make and operate buildings has contributed significantly to the climate catastrophe. AI gives us new tools there to bring about better solutions for an increasingly complicated world.

Viktor: That's right. It may well be that we receive new kinds of recommendations for more ecological management that had not occurred to us before. With all the openness to technology, however, we have to pay attention to the ecological footprint of software and, in the context of real estate, of smart components, and simply make a realistic total cost calculation that pays attention not only to CO2 avoidance in operation, but also to the resource input and the effect over the entire life cycle up to the end-of-life. A low-tech solution could make more sense here. However, I believe that AI will help us in the future, also in the context of the construction industry, to transform to more circularity. Image understanding, in combination with sensor-based data, can help us digitally map the inventory and then use AI to examine it for recyclability or reusability. This will also bring real transparency to the inventory. But not everyone in the industry will like that. Greenwashing won't work properly then. Before I talk any further, I wanted to ask you, as an architect who is also interested in our society and social issues, how you see the effect of current developments on society?

Eike: I assume that the way we look at ourselves and the world will also change. Much of what has defined our day, our week, our life will no longer be necessary. Yuval Noah Harari believes that we are being driven by a blind, technological evolution that cannot be stopped by anything or anyone. Evolution has no goals and is not an actor. So are we almost inevitably transforming into an unimaginative, AI-dependent society, and in the end, abolishing ourselves? Are we experiencing the beginning of the end of the Anthropocene? Or will we also succeed in developing a new ethic in dealing with the new technology? And to shape and use these learning machines for the benefit of all.

Viktor: I don't think evolution is blind in this area, but simply that few people have real insight in the field. Even as a computer scientist with x additional courses in AI, I don't think I have real insight here.

Eike: That's what I mean. Despite, or just because of many billions of conscious, seemingly rational decisions, something emerges, quite naturally, step by step, without one or a few having aimed for or even foreseen the result on a large scale.

Viktor: My assumption is that we are rather driven, but by technically savvy data and software engineers. Much of what one hears is not characterized by technical expertise, but often resonates with fantasy, resentment and inflated expectations. But that's not necessarily a bad thing; it encourages discourse. Who would have thought that my grandmother would ask me about AI because she read something in the newspaper? In principle, however, it is advisable to take a deeper look at AI, at least at the basics. You don't even have to program, but linear algebra and analysis are already sufficient in parts. The more factual the discourse, the better for meaningful research and development, which can ultimately lead to a more livable and ecologically more intact planet. Whereby one can also say that our planet would benefit from a little more faith in peer-reviewed science, regardless of whether it is about topics such as mobility, agriculture, energy supply or construction.

Eike: Do you think, Viktor, that the stakeholders are capable of such leaps in development? Is the real estate industry and the construction industry positioned to make these leaps in development? 

Viktor: Let me get my crystal ball (laughs). In my time as a digital transformation consultant, I've noticed that most companies are unfortunately not particularly risk-averse, nor do they invest in research & development. Materials and machinery manufacturers are much more agile. The large rest of the construction and real estate industry is unfortunately organized in a very small way, has no R&D history and is running behind. There are also hardly any formative IT departments that develop their own solutions. Hardly anyone is involved in corporate venturing either. Frederik and I have been presenting the idea of Acquirepad again and again since 2016, and all the investment and asset managers and project developers we spoke to found it interesting, but no one wanted to implement it. Now we are doing it ourselves and we are successful. That would have been possible even earlier.

Eike: We also need a lot more cooperation at this point, i.e. a large coalition of the various competence bearers from business, public institutions and science, Viktor. Everyone at the same table.

Viktor: The associations would be predestined for this, but here, too, we often come up against limits. I have experienced this all too often in recent years.

Eike: I see it similarly, Viktor. I am currently involved in the development of a software program: SOMA is intended to simplify and improve cooperation between building owners, architects, specialist planners and authorities. The program links protocols with schedules, to-do lists, invoices and makes everything transparent for those involved. And gives them more time for their actual tasks.

Who would like to join in?

Viktor: I think that's really great, Eike. That's exactly what it needs. More companies should do something like this, and so should the associations. Let's use the opportunities that technologies and changing market conditions bring together (!) in the industry. 

Eike: Thank you very much for the inspiring interview, Viktor!