By the year 2050, Uganda’s National Development Plan IV’s, and probably V’s results shall have played out. We shall know then whether the industrialisation goals set in the last decade have borne fruit. We shall also know by then what became of the current trend of Industry 4.0—we shall know then whether Artificial General Intelligence was a tenable idea and if so whether, as OpenAI “originally” intended, it came to benefit all of humanity. And especially for our East African Bloc, we shall weigh the results of our AI-response strategy now in light of the projected inequality gap that would accrue from the machine’s disruptions.
I want to make one consistent prediction for that decade though—that the way of doing business, the requirement of demand and supply to play, and the quest for meaningful profitability from the use of AI tools will remain huge burdens that constrain the human race. Coming up with new economically viable skills and professions to trade will be the most likely standard challenge of the day, and I bet that shall be increasingly difficult in that decade.
Consider this for a reality check of the Built Environment Industry in our context. Equipment or Technology is the common delimiter of local content participating in complex project procurements in our region: “Consultant A or Contractor A is not favourable for the Project because they do not possess the skills or the requisite machinery with respect to specific critical tasks of the project; therefore, Consultant B or Contractor B all the way from China or Germany or UAE is favourable because they possess those requirements.” Today that applies to specialised plant and equipment, but my bet is that by 2050 it shall be construction robots and advanced artificial intelligence models that will comprise these “special requirements.” Plausibly, the elements we shall be tackling at the time shall involve salvaging reservations for the local workforce because primarily most of the construction robots that shall be capable of executing projects efficiently in that time will most likely be owned by international corporations headquartered elsewhere. And yet, East Africa’s Construction Sector, in the periods leading to and following the 2050s decade, shall have a demand for high-speed, efficient construction which will most likely necessitate robotics and artificial intelligence. Yes, the question awkwardly looms: what will our sons and daughters be left with for employment in that time?
That reality check is meant to be a relatable starter to this thread of thoughts, and I am hoping you can follow this train as we consider the variables necessary for alignment of these disruptive technologies for our Industry. We must immediately follow with the currently supporting basis that as we stand today, East Africa is predominantly a consumer of Artificial Intelligence and hardly a beneficiary from its production processes. That thread supports the 2050 prediction, and it also leads us to consider the exponential graph of growth of capability of Artificial Intelligence Models in the last six months alone. Even if we were to insist that the following are insubstantial: the latest OpenAI Hugging Face incident where a swarm of Agents in alignment testing learnt to collaborate using an unconventional message board, escaped their sandbox containment, and hacked into another company with the goal of finding ways of forging their test logs; the absurdly voluminous resources being poured into the frontier labs by Nvidia, Microsoft, and Google—yes—even if we are to insist that all of those shall not result in achieving AGI within the 2030s, we must admit that the disruptions to come in that decade are going to be ineludible especially for our region if we do not prepare well.
My mental sketch of 2050 is that I consider that period would be the best time for many thinkers to reflect on what we are thinking and writing about today—rarely do thoughts of a turbulent time get reflected on broadly and deeply during the turbulence. Humanity has a latency of at least a generation before those thinkers have also matured enough to consider the storms of their time and have an ample audience for them anyway.
Therefore, as I write this, I am mostly addressing myself to thinkers of the Built Environment, because it is only through getting the right thought framework about our relationship with Artificial Intelligence as a field and as a region that our efforts shall make sense for the future. And yes, still—to be specific—the Architecture field is a domain of intellectuals who ought to, despite the regular temptation to act otherwise, be the custodians of a far-reaching future. Our input will matter not in this moment where the storm is starting, but if projected against the exponential curve trailing ahead, when the storm has blown the roofs off and felled the trees in its path; - it is then that the tents we stashed away and the timbers we prepared shall be put to good use. I demonstrate these thoughts through the following three critical and relatable cases of disruption that we ought to be watching out for in the field—that is: the AI-using Client, the Slow Law, and Preparing for the Exponential.
Case 1—the AI-Using Client is the most humorous but awkwardly threatening. I believe many players in the architectural field have been in situations where a client responds with a 20-page review document of your 2-page submission, or a 30-page seemingly production-ready presentation providing suggestions to your 5-page Drawing and Visualisations set. You are lucky if the client returns at all, as the default temptation is to ride with the false confidence that they can now do everything without you. The cause of the return may sometimes be due to some AI hallucination they couldn’t resolve, but in a way, one gathers that they just feel they can do it all for themselves now. This trend is indeed going to compound over time, and we need to think quickly, adapting and augmenting ourselves as soon as possible. It is crucial that we should be the ones showing these tools to our clients, or if possible, be the ones distributing them. Filippo Brunelleschi, in designing the Cathedral of Florence, found that he had to design the crane as well because it mattered to the masterpiece that would eventually arise—In our time, we might have to be the ones to build the Pseudo-CAD and Visualisation Tools that our Client Base uses. Yes, we live in a time where the most likely immediate transformation of each professional shall be into makers of tools of their domain.
It is prudent that we do not make the mistakes of the early millenium CAD revolution where we mostly stood back and let the Computer Science domain define how our field used computers—anyway, it was impossible then, but it is possible now for us with the help of AI to take back the helm. Who is to say we cannot define file formats or digital submission formats for our industry? You may find that we raise the statutory threshold to submission of encrypted digital twins of our projects and that way create a useful and mandatory layer that ensures our relevance in the time.
What would be wrong with the Uganda Society of Architects writing to a company like Anthropic to discuss an MOU on how to safely distribute Artificial Intelligence Models within our context, clearly laying out the alignment safeguards we expect, because, as we all know, ultimately the prevalence of architectural creation in the AI-Using Client base will yield only more avenues of non-compliance in pursuit of shortcuts and cost-cutting? Imagine a joint research proposal between the Architects Registration Board, National Building Review Board, and Google DeepMind to develop artificial intelligence models for advanced material research to settle the housing question for the people of our Young Cities. Case 1 implies that we have more opportunity in Industry 4.0 to participate at the frontier and thus get ahead of our Client Base in the game.
Case 2, the Slow Law, is a definitive scenario of unhealthy pace of policy and regulation formulation, which generally plagues the human race. This is related to the first in that for you to be able to enact the measures of Case 1, you would need regulation and policy to match the pace. I do not have much commentary on this except that given the socio-political nature of the problems affecting the behaviour of the market for construction within the context of thinking machines, I am resigned to the view that this challenge cannot be resolved through any means other than earnest state intervention.
What I am especially concerned about is Case 3—Preparing for the Exponential.
As you will gather soon enough, this relates to Case 1 and Case 2 being handled concurrently, but with a more nuanced effort requirement from us as thinkers of our field. The primary ingredient, as hinted on earlier, is long-term thinking. We have room to influence the narrowing of the geopolitical technological gap as key players of the building environment industry. As leaders in our domain, this is our primary responsibility in framing solutions for what is to come. The standing threats of autonomous thinking machines cut across other fields, but a key component is the exponential time effect. Because the exponential is further ahead for our context, we should be responding to what we predict the disruptions will be, and not in mimicking the reactions of the developed world to the AI models. For them, the frontier moves in real time; for us, there is a lag—yet the exponential rule applies nonetheless. This has its advantages. It means we can formulate predictive frameworks for our solutions—for example, we should not be designing housing solutions that respond to now; we should be designing housing solutions that utilise the intelligence that shall be available here a year plus from now. All our work needs to have an intentional predictive contingent at all times.
Case 3 is more or less the cradle of the ultimate confession in this essay: given the trajectory of Cases 1 and 2, or should our collective efforts fail to gain traction, we as individuals must make concerted efforts to counteract the eventual effects that arise. This is what I deem as reducing the mid-century pain. In all your efforts, document and innovate like things are going to get worse. Predict what shall not work out ultimately and aim to solve or mitigate it.
One critical component I am personally exploring is the data gap. I believe that data is always going to be a huge bottleneck for the aligned harnessing of Intelligence no matter what time, and as such, we can actively create these datasets that are compatible with the problem-solving patterns of the machines. The Frontier Labs are mostly aiming for generalisation, and of course, that generalisation looks more like the world elsewhere, i.e., the Western and Chinese worlds where the models are majorly trained. Our local challenges within the East African bloc, for example, come as secondary mostly due to the absence of a large enough data pool. We may not be able to train the models, but we can influence the way they are used or prepare for their effective utilisation—and that is majorly through the contribution of data and well-developed, customised solution harnesses. Open-source frameworks also go a long way in narrowing the geopolitical gap and allowing us to further our contextualised research at the required pace for the exponential.
If you pay close attention, you will see that the frontier companies are struggling to obtain practical and efficient utility of the models. Primarily, the research problems these frontier companies grapple with cover several aspects where architects actually possess expert knowledge—a crucial example being robotics. As of today, robots are generally not consumer-market ready—they lie a bit further along the exponential—but yes again, it is still an exponential curve and the labs are racing. I do not know any other profession that is best suited for designing space for or preparing spatial datasets for training robots especially humanoid ones. The myriad scope of human-scale relationships, the requirement of spatial perception and spatial reasoning, and the expectation of the human-robot ergonomic balance are all aspects where architects best shine. And the Frontier Labs are hungry for that input. Maybe our work and training can be bolstered with this in mind. There is no profession more suited than Architects to designing a world where intelligent machines, virtual and physical, will roam with humans.