A trip of a lifetime

Every day is day one. Day one. What is that supposed to mean, you might wonder? It’s great piece of insight offered by Teppo Kuisma, COO of Digitalist San Francisco, a tower of a guy with a big smile on his face. The insight is that we cannot change the past, but we have the power to affect what will come. While this might be useful to keep in mind in the hectic start-up scene in the Bay Area, it also applies to the IT business in general.

My journey to Silicon Valley started already a half a year ago. That was when Bilot and Louhia arranged a reverse hackathon around AI. The winning prize was a trip to the Silicon Valley, to meet with both start-ups and established major players pushing the limits of AI. As a member of the winning team, bringing intelligence to dairy farming with Raisioagro, I got to participate in this amazing trip. In addition to the impactful official agenda, we also got to experience the wonders and nature in Silicon Valley Palo Alto and scenic San Francisco city.

Three, two, one… lift-off!

As the departure came closer, I started to have serious doubts about travelling; what could a long-term academic like myself gain from such a business trip? How wrong I was! While I was mostly interested in the AI aspect of the visits, and those were at best jaw-dropping, the stories of entrepreneurs behind successful start-ups, such as Idean, were truly fascinating. Making or breaking in this extremely competitive set-up can be a matter of luck and good timing, but success does not come without high commitment and the ability to adapt.

On Monday morning, the agenda took us first to Singularity University—at NASA Research Park, an organization that aims at tackling the world’s biggest challenges by leveraging the possibilities of emerging new technologies. The discussions we had there made me realise that technology—not just AI, but all tech—is developing at an exponential rate. I was actually quite surprised that I had not thought about this earlier, but now I can see that the ramifications of the pace of development are quite mind-blowing. For one thing, we are gradually losing the ability to understand the technology we are generating, which resonates with the famous quote by Arthur C. Clarke: “Any sufficiently advanced technology is indistinguishable from magic”. I can tell that this first visit really set the scene for the coming meetings; If you think big, everything is possible—even the unimaginable.

Salesforce headquarters in San Francisco (the cylinder in the centre). Spectacular view from up there.

Highs & lows

The next highlight of the trip was, somewhat surprisingly, the visit to Facebook. One thing that became quite clear is that many large service providers, such as Facebook, Amazon, and Netflix, do plenty of machine learning research and application that is neither part of their main service nor visible to the user. Let me give you an example: While about half the internet users are on Facebook, it seems that the best way to increase their user base is to improve internet access generally and to bring free internet to places where access is very limited. This involves, for example, building a grid of autonomous planes that can be used to generate a mobile network to places with no relevant infrastructure, such as Africa. In their normal operations, Facebook uses AI to filter up to 2 million inappropriate accounts a day. They also have active development, which is open-sourced (https://developers.facebook.com/products#open-source).

Slide at Google campus. We were not allowed to try it out due to, you know, insurance issues.

Not all meetings were that spectacular, though. I had high expectations about what Google might have to offer. While we had CTO Ron Bodkin (responsible of applied AI) talking to us the rapid development of computer vision, speech analysis, translation, natural language processing, recommendations, and predictive responses in Gmail, we were left hungry for future visions. Given that Google has been the forerunner of AI development in the recent years, there must be plenty of those on the drawing board. Another disappointment was our visit to HP labs, which didn’t really provide any useful insight. While a guided tour to the historic offices of David Packard and Bill Hewlett was interesting, it left a feeling that the company is mostly looking into its great past rather than to the future.

History was made here. The “HP garage” (at the end of the driveway) in Palo Alto.
ML models require labelled data for training, but such raw material is not readily available. Figure eight is an interesting company from San Francisco that focuses on providing a service that lets users to annotate and label unstructured data, using a vast community of contributors to do the job.

This brings me back to the beginning of this short report; every day is day one. Why is this advice helpful for an IT consultant like myself? The shortish answer is that the rapidly developing technology creates an increasing demand for updating one’s own skills, as well as adopting new tools and new ways of thinking. In this game old experience can be useful, but it is not enough for remaining competitive. In addition to Teppo’s piece of advice, we returned home with our heads full of ideas, new potential partners, plenty of great pictures, heavy jetlag, and a dream of returning to the Valley.

The team and hosts at SAP.io meeting, San Francisco.