From Saddle Making to Why We Shouldn’t Fear the Rise of the Machines
Most people who asked about our underlying tech focused on how the physical saddle shells are made. We, on the other hand, are much more interested in how the 3d model of the shells are generated. By automating this step, we let machines handle the bits unique to each person (i.e. the customization aspect). Machines can do this much more quickly, at significantly lower costs.
Using machines to help realize mass customization doesn’t mean that the human designer is made redundant. Instead of designing individual saddles, (s)he works on model parameterization: the designer determines what the relevant model parameters are and how they impact the shape of the model. Machines simply scale, or amplify, this human effort.
Scaling Human Effort using Machines
Examples of scaling are commonly found in our lives today, and is present whenever there is a hierarchy. In companies, the boss makes a decision, and his crew executes his plan. In the entertainment world, the efforts of actors and actresses are broadcast in cinemas and via Internet streaming reaching a large audience. In the software world, a program written by an engineer can be run on hundreds of thousands of servers in data-centers around the world with a click of a button.
Instead of thinking about machines being our overlords, we should think of them as boosting us up higher in the hierarchy. Instead of worrying about them taking over our jobs, we should view them as tools making us significantly more productive. We recently had a series of winter storms which caused a number of landslides to block our favorite riding roads. If the road maintenance crew is complemented by machines, our roads can be cleared much quicker, and we can get back to enjoying riding much earlier.
Finally, being higher up in the hierarchy is usually not all THAT bad. After all, many people still want to have a career and climb the ladder.
If we decide not to fear machines, but utilize them to our benefit, what’s next?
Two relevant questions at this stage are: how do we determine if a task should be done by us instead of machines, and what sort of skills do we need?
Specialization & Instruction
The typical answer to the first question is the level of creativity associated with a task. To invent a new game, we need a human. Machines can beat the best human chess players, but they cannot invent a new board game. On the other hand, machines are more suited to repetitive tasks that require precision. Even Foxconn, the manufacturer of our iPhones, is actively automating their factories. This assignment of specific tasks to different entities is not new: specialization improves overall efficiency and productivity in our society today, it is why we have farmers, teachers, plumbers, computer scientists, chefs etc.
With regards to the second question: the required skill set is somewhat different from that of a hierarchy consisting solely of humans. In a human-only hierarchy, managers have to understand what their subordinates’ tasks are, and to instruct/correct any issues encountered. In addition, there is the human-human aspect of interaction which needs to be managed. For the human-machine hierarchy, the need to understand machines’ tasks is present, but the human-human interaction aspect is unnecessary when communicating with machines. The biggest difference is the way we tell the machines what to do, what they are doing wrong, and how to correct their mistakes.
Today, we instruct machines primarily via computer programs. For instance, our online saddle design service is programmed using Python. The ability to program is typically associated with a rather steep learning curve. To accelerate the permeation of machines into more aspects of our lives, we may need a more human-centric way of interacting with them. In the case of the road maintenance crew, it’ll be great if we can simply tell the machines: “The ground here is still too soft, we can tell from the way it sinks when we put weight on it. We need to reinforce it first.” and have that be understood and acted upon by them.
The Human-Machine Communication Barrier
But, there’s a lot of effort today to get machines to even understand what we’re saying. In the last episode of this season’s The Grand Tour, Clarkson’s voice message to May was converted into a hilarious written version by his car. We keep trying, and trying, to get machines to understand human language, when they clearly aren’t very good at it. Or get good at it only after we’ve thrown massive amounts of resources into solving the problem. Would it be a more efficient use of resources for us to understand the machines instead, which requires training in programming languages? Or perhaps it is more likely that some sort of middle ground between human and machine language would be more appropriate?
For Silicon Valley Fans
Let’s all try and ‘make the world a better place’, via task-based hierarchical structuring of humans and machines, and realization of efficient and effective human-machine communication protocols.
Interested in ‘normal’ saddle-related articles? We have those at our blog page.