There is this artist Berndnaut Smilde who makes these amazing clouds inside of buildings. He’s been making these clouds for over a decade now. He’s been quoted as saying, “I cannot really control the cloud.” Meaning that these synthetic clouds that he made only physically last for a few seconds. It’s the photo of the cloud that has the timeless impact. But we do have control of the cloud as in data and cloud computing through AI for IT operations, which we are calling AIOps that has the ability to persist and span across time reaching into the past and showing us possible futures.
Relating to art, Lila Tretikov, a deputy CTO at Microsoft, explains that art and design is one of the best ways to think creatively and imaginatively, and then with the use of science and technology, that vision can become a reality. She says that it is very important for [scientists] to keep interacting, working with and co-creating with artists because artists can help to ideate and envision tomorrow. Right now, we often serve technology, rather than technology serving us. We have an opportunity now, especially with AI, to truly have technology assist us, rather than us drive technology. We have the ability to create AI in way that is assistive, and helpful to us, as opposed to destructive.
Demis Hassabis of DeepMind said a lot of the new breakthroughs that are going to come in the next decades are going to come from the intersection of different subject areas where there’ll be some new connection that is found between what were seemingly disparate areas. Personally, I think art and AI is an interesting cross-disciplinary inquiry drawing on my formal training in art and decades in tech.
Oriol Vinyals of Deepmind says science fiction has prepared society to start thinking reasonably about certain topics, and for the inevitable research. There are many important challenges and topics that come with building an intelligent system. We’re never going to be fully ready unless we talk about this and we start expanding the people we talk to, to not only include scientific researchers. There are more interdisciplinary groups forming to start asking and really working with researchers on questions because I’m assuming an interdisciplinary topic is not initially where the passion lies when future scientists are getting their PhDs.
Oriol has seen an increased level of maturity in the conferences that deal with AI compared to five years ago from now. The number of workshops has changed so much it is impressive to see all the topics, including ethics, that come to the surface, which is great and if we are too early, clearly, it’s fine because it’s a big field and there’s lots of people with lots of interest that will make progress and Oriol doesn’t believe we’re too late.
So, there is value in cross-disciplinary work between scientists and sci-fi writers. The writers are important because they focus on thinking outside of the limitations of what is scientifically possible at the moment. Along with writers, visual artists provide the vision of what the future could look like, and there is also a need for scientists to ideate and envision tomorrow. The above is Brian Aldiss Hothouse written in 1962 envisioning a future for Earth where plants dominate half the world. This work can be seen as a prediction from the 1960’s for the climate crisis we’re having now.
In this illustration that I created, I was really inspired by the AIoT work the local Seattle company Koidra, who by the way presented in last year’s KDD AIoT workshop, was doing with smart greenhouses. My thought was to take the greenhouse to scale, with the vision of an endless greenhouse equipped with sensors, and portraying AIoT in drones detecting, diagnosing, and predicting the future of crops. In a way, we are making Aldiss’ imagination a reality.
Because I deal with AI and art, it would seem that I would have an opinion about Open AI and Dall-E. Which I do. I took the previous hero image and ran it through Dall-e to get the above, which, interestingly, have a very painterly and more abstract look. More of a raw material with a human touch look, than something that was purely created in blender and photoshop as in that previous image. Although aesthetically more appealing to my eye, the utility of the illustrative function has been lost to something that at the most is suggestive of a drone flying over plants. The last one is particularly interesting in that it looks like it gave the drone its own gravity warping the grid.
Back to blender and photoshop, the above is an image conveying the idea of a self-driving car in a night scene, with the suggestion of a city on the horizon. So just using a wireframe of the car and the representation of sensors and the range of the sensors consisting of binary data. With self-driving cars, they bring in so many different areas of our culture, politics, humanity, and the emergence of typical ethical dilemmas that show the importance of ethics in the development of technologies. For example, you probably know the example of an autonomous car with broken brakes going at full speed towards two pedestrians. Both pedestrians are in danger, but by deviating a little, one can probably be saved. These ethical dilemmas become more prevalent as AI becomes more general. It should be noted that we are part of a company that is actually committed to the advancement of AI driven by ethical principles that put people first.
Here is my illustration for AIOps relating to predictive maintenance showing how across time, we can detect, diagnose, and predict. You can see the car identified in the back as a past state with its history of data, then in the middle, in the present we can diagnose the properties of its existence right now, for example the tire condition with the exploded view in lime green showing the data points for each wheel, and then the predicted future state based on what we know from the current state and the historical analysis. To reinforce the idea of time, I literally put clocks on the floor. I also overlaid a neural mesh across the floor as well as code on the left.
What I understand of what smart cities are is where there is effective integration of physical, digital and human systems to deliver a sustainable, prosperous, and inclusive future for the people who live and work in those cities. Smart cities optimize the use of technology in the design and operation of infrastructure and buildings in a way that meets the current and future needs of their citizens. This is an illustration for MLSys2022 Cloud Intelligence / AIOps Workshop. This is a generative work that shows, again an infinitely scalable landscape, this time with buildings which visually adheres to the principle of design using variety within repetition. You can see the commonality of the buildings on the x and y axis and then the variety being the height, or the z axis. And then having the viewpoint from above, gives a sense of control, or at least an overview, giving a better understanding of the city as an entity in itself.
And finally, here’s an illustration of weather prediction with remote weather sensing stations. There are public bus stops that have these sensors attached to them that gather data. And then I took a photo of one and built on the photo by showing clouds with overlays of data, a neural mesh, code, and topography. And incidentally, this code is open source used for weather prediction.
***
#AIOps #AI #art
