Video Script (Video Length: 6:50)
Over the next few years, Collaborative Intelligence (abbreviated as CI) will have a significant impact on many areas of your business. This video series will provide a brief primer on this topic for business executives and other business leaders.
This is the 2nd video in the series.
There are three categories of applications for Collaborative intelligence. While there is some overlap among them, distinctions arise from differing goals and the roles that humans and machines assume in each category.
The first is redesigning work in ways that improve human performance. Intelligent machines enable humans to do what they’re doing today better, faster, and less expensively. But they also enable humans to do things that are beyond human capacity and to do entirely new things.
For example, one of the things that intelligent machines excel at is ingesting large amounts of data, synthesizing it, and finding patterns within it. Thus, you could ask an artificial intelligence application to:
read 1000 journal articles on supply chain management,
identify the five most important trends that may emerge in the next three years, and
highlight the three best frameworks for making sense of these trends.
On a pre-trained model, this would take, at most, an hour or two. While a human could do these same things, it would take them 2-3 months, if they read all the articles. The AI enables the human to complete the work much more quickly and less expensively.
The second category of applications is Optimized Task Sharing. In this type of application, workflows are redesigned, with specific task completion steps assigned to either a human or an intelligent machine, based on which is best suited for the task.
One non-business example is Freestyle Chess, which allows human players to use computerized chess engines and other resources during a game. Such an approach assigns the task of calculating the results of moves, countermoves, and probable outcomes to the machine, while assigning tasks such as strategy design and execution and the assessment of the opponent to the human. Research has found that the combination of humans and chess computers outperform humans or chess computers when either is working alone.
The third category of application assigns human oversight to the work of intelligent machines.
For example, in quantitative finance, intelligent machines examine as many as 10,000 prospective investments, analyzing and filtering them to ensure fit with the portfolio managers’ selection criteria. The machines then invest in those that fit the criteria, while also adhering to other constraints, such as those related to risk management.
Prior to going live, humans will simulate the investments and monitor the portfolio’s performance to ensure that the real-time performance is similar to the backtested performance, while also ensuring that all risk management policies are adhered to.
After some time, usually six months, if the algorithms work well, they go live.
ChatGPT and other Generative AI technologies have been getting a lot of attention lately. But Collaborative Intelligence applications have been helping companies improve their performance for several years.
I provided generic examples of each category of application. I will now present specific examples.
BMW is using a technology called Digital Twins to reduce the cost and time associated with designing or reconfiguring their manufacturing facilities.
A digital twin is a digital copy of a physical process, physical system, or physical entity. It mimics the physics and other physical characteristics of its twin in software.
It enables users to modify the digital twin to optimize its performance or examine the impact that prospective changes will have on the physical twin. Once an improved or modified configuration that meets all performance criteria is found, the changes can be implemented in the physical world.
And that's what BMW is doing. It is using NVIDIA’s Omniverse platform to enable its globally-distributed design teams to work together in real time to propose and understand the impact of prospective changes to its manufacturing lines, including impacts on worker ergonomics and safety.
Once everybody agrees on the best way to do things, they implement them in the factory. Using a digital twin enables BMW to make changes more quickly, less expensively, and with fewer mistakes.
An example of using task sharing to improve product development is Autodesk’s DreamSketch and Project Dreamcatcher platforms, which, like ChatGPT, are Generative AI technologies. The platform allows designers to input hand-drawn sketches of ideas and a set of design constraints, and the platform will generate as many as a thousand prospective designs, all of which meet the specified criteria and do not violate any stated constraints.
For example, a designer can use a tablet to draw their idea for a new office chair and enter constraints such as it having to be:
constructed of aluminum, faux leather, foam, and rubber only,
must support at least 350 pounds, and
be no more than 18.5” wide.
The platform will generate many designs. Some will be similar to the sketch; some will be very different, taking approaches that the designer had not considered. All will fit within the specified constraints.
The designer can select those that are most interesting, make tweaks using their design tools, and have the platform design variations on these themes, iterating until the designer has selected the final design.
This approach to task sharing reduces design costs, speeds time to market, and often results in more aesthetically pleasing or artistic designs that bolster the company’s brand.
Finally, one example of human oversight of intelligent machines is the work being done at Ten63 Therapeutics, which is using their proprietary computational chemistry platform to generate and screen trillions of molecules in order to discover new drugs that have the highest probability of overcoming resistance and of maintaining efficacy in response to the most likely mutations.
Without getting into the technical details, their platform has enabled them to develop several promising candidate drugs. Currently, they have two cancer treatments in their pipeline, both of which have shown much better performance than existing treatments.
These are just a few of the many examples of how companies are using Collaborative Intelligence to improve their performance.
How will your company use CI to improve its performance?
Share this post