It’s been a while since a wrote on my personal blog. Recently, I’ve been mainly writing for Mainly.AI. Do you know my writing style well enough to figure out which of the blog posts are written by me? I can give you a hint: I love bullets. I always think in terms of bullets. After bullets come pictures. After pictures comes text. Let me tell you more about bullets.
A bullet is typically a self-contained piece of information one can refer to. And I love references.
When I read plain text that does not contain bullets i normally try to decompose it in logical blocks and organise them in my head, which helps me to remember.
Am I a robot? I love machine-readable structured text. Structured text is also so much better for scientific articles when it comes to finding right pieces of information automatically.
Shooting from the hip
I’m not the person who overthinks things. Or am I? I often get to hear that I am assertive. I take it as a fine compliment. At times I do however have a hard time choosing between Pareto-optimal solutions and my brain goes at full speed even though I know in theory that it does not matter which one I choose.
I do indeed shoot from the hip sometimes. Crazy ideas, assumptions, hypotheses – I love them. But only in a safe experimentation environment and with all those constructive critics around me who are not afraid to object. Brainstorming at its best is a like a scientific experiment.
Technology is simple, people are difficult. People create a piece knowledge, like this one: Coronavirus disease (COVID-19) advice for the public, which also has a timing aspect to it. This piece of knowledge immediately starts spreading and transforming on the way. Knowledge is there to be spread, of course, but there are different ways of doing it. The way I just did it myself, by linking to the original piece of knowledge, does not give me a piece of that spotlight. In a search of a piece of spotlight, people start para-phrasing the original piece of information, picking out pieces, adding own views and passing it on. This leads to a plethora of information pieces out there, with no possibility of backtracking to the original knowledge object.
What’s the mechanism of retrieving the ground truth, that initial knowledge object provided by empirical evidence? An answer to this is linked data. Instead of copying and passing on a piece of knowledge we send a reference to it. This is why I am against sending files via mail – you never know which version of the file you are getting. If instead we only share pointers to knowledge objects we can choose to always get the latest. The knowledge object can by itself evolve as well but keep track of the changes and detect if anyone has tempered with it.
To complicate it further, people, including myself, love detecting patterns in pieces of information, combining knowledge objects together and inferring new pieces of knowledge. We need to make sure we can back-track this chains of inferencing to original facts and ground truth, in line with what Hans Rosling said in Factfulness. A tiny tweak in a piece of information along the chain of reasoning may lead to an incorrect decision in the end of the reasoning chain.
The tiny tweaks may be intentional and unintentional. A minor variation of the ground truth or an error in the reasoning chain may lead to wrong decisions being taken at the end of the reasoning process. When this process concerns life and well-being of people, business-critical decision-making, or societal challenges, it needs to adhere to certain principles:
Data should never be copied. Send pointers to data, not the copy.
Traceability and explainability in decision-making needs to be in place.
In a search for optimal decision, don’t experiment on a live system without boundary conditions.
Back-tracking should be possible.
Mechanisms for resolving conflicts should be in place.
Mechanisms for detecting tweaks in data should be in place.
Mechanisms for reversing decisions should be in place.
I am going through yet another leadership training program and yet again I get the same verdict: goal-oriented, structured, achievement-driven, knowledge-driven. I whisper into my colleague’s ear: “I am not a people person”. He whispers back: “then maybe you don’t have the right job”.
The fun fact is that I find a lot of people who are like me. So, I started to analyze the strengths of my leadership team as a whole and came up with; warm, innovative, fun, and some great people leadership skills. So luckily, together we represent all the different but important pieces that we need for this organization.
It got me thinking, last year I wrote a blog post about playing solitaire. This time it’s different– It’s all about putting together a jigsaw puzzle.
For instance, my leadership team comes to me and says: “Elena, we need to have regular all-employee meetings, and no, it’s not OK for you to join online, be present, visible, receptive” – exactly the push I need. I am thankful to have support from my team in various dimensions – we complement each other as pieces of a jigsaw puzzle. Each team member comes with a set of special skills, and together we are complete and efficient.
Each piece of our jigsaw puzzle is unique – we have different backgrounds, skill sets, preferences and passions. My unit is a self-going agent within a company close to 100,000 employees representing all its values. Our shape is perfect for our mission, to be the driver of artificial intelligence technology leadership for Ericsson. But if you zoom out, we are a piece of a bigger puzzle.
Let’s now zoom back in again. It’s not just our unit that has a shape. Each person has a shape as well. A T-shaped person has in-depth knowledge in one field (vertical bar of T) and a broad knowledge of an application domain (horizontal bar of T). For example, a statistician specialized in political systems. Or, a politician who knows statistics. It is important that the two bars are connected. For instance, if you are really good at solving differential equations and know a lot about French porcelain of 17th century, then you cannot automatically call yourself a T-shaped person, as they are not really connected the same way.
And similarly to T-shape, there is a Pi-shape with one more leg of in-depth knowledge, and an E-shape… you get the point. In general I would say, the more “bars” you have as a person the more interesting you are. Of course, and in case you choose to be I-shaped, you can still be endlessly interesting for people who are into the same field, but you run a risk of being seen as ‘just a geek’ by everyone else.
Personally, I have been working hard on diversifying my professional profile (or adding more bars to my shape) while keeping it all connected. I gladly take opportunities of diving into new areas and shifting my focus from old areas. In the long run this strategy should bring me to the shape of a Swiss army knife. There are plenty of “bars” of different shapes and directions, and they are all connected, and one can actually choose which parts of the knife to fold in or out. Or, if we were to tweak the rules of the jigsaw puzzle, the more skills you have, the better you fit in with the other pieces.
Still, having all these different skills is one thing and enjoying using them is another. Doing a good job is not necessarily the same thing as having fun at work while doing a good job. Being a fan of multi-parametric optimization problems, I aim at maximizing the fun, so that we all get to use our favorite skills. So forgive me for not being a people person, luckily I have my team to compensate for that. And next time you find me hiding in a corner during a mingle, please come by and say: I read your blog post!
Do you know what’s the best strategy for playing solitaire? Balance your piles of cards.
When you have a choice between piles to reveal the next down-facing card, you should pick from the largest pile. It has been a while since I played solitaire, but the rule got stuck in my head. We all have many different piles in our lives, formed by our jobs, families, and hobbies. To succeed in the long run, we need to pay attention to all the piles, and specifically to the bigger ones. Every time we have a deadline at work, we give that pile more attention, which is ok for a while as long as we don’t forget to shift the balance later because we want to win the whole game, not just empty one of the piles.
Now, imagine that our piles represent blocks of different opinions. It may seem tempting to only work with opinions resembling your own. That way you may seem efficient in the short run but will be doomed in the long run because diversity is important for success. Diverse teams create the most innovative ideas and solutions. When you run an organization, a project or a meeting, don’t forget all the different perspectives, otherwise, in the long run, you may end up in a local minimum.
I love my diverse team at Ericsson Research. AI Research is composed of researchers based in Sweden, India, USA, Brazil, and Hungary. We have forgotten why physical co-location was so important and instead do virtual co-location. Working on the same data sets, within the same environments through the same tools and on the same projects is way more important than physical closeness. If only we could do something about the time difference…
On the other hand, when we in Europe come to work, we can build on the results from our Indian colleagues, and when we go home, we hand over to our colleagues in The Americas, so that cross-continental projects can deliver results around the clock. Diversity has a great social aspect to it as well. Our diversity is something that we all have in common, and something that we gladly discuss with each other.
Now to my biggest passion – technology. Working with Artificial Intelligence is fun – it’s broad and deep and can be applied to so many things. And one of the things that fascinate me most is that AI technology is inspired by humans and nature. This means that whatever humans found to be successful in their lives and in evolutionary processes can be used when creating new algorithms. Diversity, inclusion, balance, and flexibility are very important here as well, with respect to data and knowledge, and diverse, organizations are for sure better equipped for creating responsible algorithms. In the era of big data, let’s make sure we don’t discriminate the small data.
January swished by and it’s time to follow up on the progress of new year’s resolutions.
Books ✔. I have accidentally read four. The lovely app I am using for streaming audiobooks is so good at serving me just right things, making me curious and getting me hooked.
Exercise. I managed to achieve 1.75 per week on average and that wasn’t easy.
Blogging. I have only written three this month (one internal and two external). I have plenty of ideas and not enough time discipline to put them on paper. You know these activities without deadline. Can someone force me please?
Teach Paul Junior to ski and to swim ✔. Planned.
Wear heels. Improvement area.
Spend more time with my parents ✔. I have trips to Saint Petersburg planned in March, May and June and July ❤
How can i improve? Read or write while exercising on high heels?
Find a woman that helps you with the cleaning and the chores,
Find a woman that is a good cook,
Find a woman that you can trust and share your feelings with,
Find a woman that enjoys making love to you,
Last and the most important thing is that these four women should never meet.“
So, I am writing an unusual job ad:
I am looking for an assistant who can do administrative tasks (but rather wouldn’t).
I am looking for a hacker who loves automation.
I am looking for an innovator capable of thinking outside the box.
I am looking for a people person who is good at understanding people’s needs.
And the most important thing is that it should be the same person.
In other words, I am looking for a person who will be able to challenge us in any task we do and see if we can become more efficient without compromising on integrity, regulations and principles. Let me tell you about some of these tasks:
Expense reporting. According to regulations, each receipt needs to be scanned, attached to the expense report, and then the whole report has to be printed and submitted. Can we optimise things here? For example, I always prefer to get receipts in digital form, so that i can immediately forward them to my assistant. Or, maybe they can automatically go to my assistant or a special inbox with my receipts? Or should we integrate with SAP directly and skip sending receipts to inboxes? (note: Uber offers this functionality already). Can we automate report printing and posting as well?
Purchasing. I have a simple rule: I delegate to my assistant to approve all purchases that are obviously motivated and are under a certain amount. These are 95% of all purchases. This sounds like a simple script to me. Don’t tell my employees.
Time reporting. Some lucky people like myself do not need to do time reporting. But most of our employees face a complex project structure. For example, when you work on EU projects, you need to track all the hours, and report them accordingly. If my time is split between five projects, can I please have 5 buttons on my screen and click on them in accordance with the project I am working with? Or, can we let an AI figure out which project someone is working on judging from the context?
Monthly reporting. I have a feeling that most of the people find monthly reporting boring when it in fact should be joyful and an opportunity to tell your colleagues about your progress. Let’s make it joyful. Let’s forbid sending snippets of text via mail. Let’s forbid copy-pasting. Let people write about things they are proud of, or things that they believe need attention. Let us use our favourite platforms for that, and don’t force people to do it once per month. And every piece of news should be hash-tagged so that later on anyone could easily create an on-demand ad-hoc report. My boss should be able to check out all achievements of my organisation towards a certain stakeholder within a certain time interval, for example.
Keeping track of employee inflow-outflow. This sounds easy, but it’s not. For normal employees we need to track the status – empty position – job offered – contract signed – started. Not all of these states are reflected in our HR system but we still need to keep track. We also have many different kinds of interns – summer interns, master thesis workers, PhD students, postdocs, some with salary from our company, some without.
Employee onboarding. When you join our company you get an account, email address, computer, phone, desk, screen, chair, etc which is a pretty generic starter kit. Of course you can choose which phone you would like to have and which computer you would like to have. I can think of a simple web form for all new employees where the computer/phone model gets chosen, and an automatic purchase gets triggered.
Scheduling meetings. I start by saying that I love meetings. It’s probably the first time you heard anyone say that. The trick is that I became very picky with which meetings I attend. I simply choose the ones that I love – content-heavy and action-oriented, aiming at creating impact.
And anything else that you may find inefficient – challenge me!
The skills: hacking, scripting, SAP, Sharepoint, RPA. You come with an open mind and we give you access to our tools. You will also get this hoodie from me personally (used, by me, during my postdoc time in Paris).
Few hours left of 2019. Time for the summary. To start with, I thought i would check in on my new years resolutions for 2019. I did not manage to hold all of them, so i still do have something to work on. Here’s a selection:
Keep my house clean – kinda OK. At least the awareness is there now.
At least one internal and one external blog post per month – Fail. I love to do this and always have plenty of things to write about but…
Workout at least twice per week – kinda OK. We got our running routines on weekends back and it feels great. Kudos to Anders Hansen!
Kilos lost -2 (c) Bridget Jones. But i heard that curves are back in fashion.
Read more books. I am a slow reader and never read with my eyes any more. This year I read 12 books, see my book club.
Looking back at my 2019, it feels like an OK year, nothing extraordinary. Here are my highligths:
Became a member of the board of directors of RISE (Research Institutes of Sweden).
Just a couple of hours left from my visit to Tokyo as part of the Royal Academy of Engineering Sciences (IVA) Innovation Leadership Program’s (IVA15) delegation. Japan is always interesting, in so many aspects. It’s the country with the longest life length in the world. The country always scoring among the highest in the world in innovation, with its high ranked universities and strong traditions.
It’s been an intensive week with 15+ visits to companies, universities, research institutes and authorities.
Human-centricity has been
lifted in every presentation we’ve seen. Perfectionism in craftmanship has
always been key component of the Japanese tradition, which is reflected in the
quality of Japanese food, hand-made goods, and the precision of manufacturing
processes. Respect for each other is another key component, reflected in all human
behaviors. A short crash course by our guide from Business Sweden on how to
dress, greet and sit, and where not to put the business cards that you receive
was valuable. Still, I managed to break the social norms a couple of times by
having my legs crossed when sitting.
There are many more jobs
than job takers. People work very hard and taking full vacation is considered
bad style. There are several ways to improve the situation – guest workers from
other countries, encouraging more women to work. Artificial Intelligence is
considered to be the game changer and life saver. AI is at the core of Japan’s
vision of Society 5.0/5.1. Opposite to how the discussions go in Sweden when people
worry about the effects of automation and AI, people is Japan look forward applying
AI to automate jobs. “If you lose your job in Sweden you can come and work in Japan”,
one of them said. And for the first time I saw that the authorities are looking
not only at replacing that low-end, repetitive, dangerous and boring jobs that no
one wants to have but also the high-profile jobs such as directors and
strategists. Agree completely – when companies are fully digitalized they can
just as well use AI as a managing director. As long as it works together with a
human board of directors.
Every time AI was mentioned it had a human in the center – “human-centric AI”.
Great concept for many reasons,
including the comforting one – “AI is my trusted partner, but I’m in control”. Quite
some work still remains to ensure that we humans are in control, which is also something
that we researchers like – stay tuned for upcoming papers on that matter.
Toyota is a role model for
production system design with its definition of lean production and the Toyota Production
System (TPS). We saw it in action and it’s amazing. TPS adheres to a number of
principles and the objective function is to minimize “waste”. The definition of
waste can vary and a simplified version means time waste. It can include more parameters
such as sustainability impacts or material waste. Many productions plants (such
as Ericsson and Scania) have implemented their versions of TPS. An immediate
thought of an AI researcher: the production plant planning problem can be automated,
since it’s very well defined.
Sony Computer Science Lab
A company under Sony’s
wings with some 20-30 crazy scientists on a mission to change the world. Can AI
win a Nobel prize? Can AI become a Michelin star chef? And researchers’ KPIs
didn’t only include papers, patents and industrializations but also impacts in art.
One exhibition showed robots developing their own language. For any object they
detected they mapped a sound and then could communicate through these sounds
with each other. In the end they were chattering in their own “natural language”
about the visitors of the exhibition. This is as such not ground-breaking nor
directly applicable in industry but very cool and makes AI understandable for
those not in the field. It is an illustration of a collaborative learning
process, which has a big educational value as well as an artistic component.
All Sci-Fi lovers were mind-blown. Others too, I believe. A lot can be said about the company and the cool CEO who took us through the history, development and successes. This was truly human-centric. Similar to Steve Jobs, the CEO of Cyberdyne, Prof Yoshiyuki Sankai, once had a dream. He wanted to help people and address the potential problems of the aging population of Japan. He built a system that non-intrusively reads bio-electric intentional signals of a human, amplifies these signals, actuates the robotic part of the solution – a cyborg-like exoskeleton, and closes the loop by sending the signals of performed movement back to the brain. Applications included patients with spinal injuries who learned to control the exoskeleton with their brain and later restore the function of their bodies without using an exoskeleton. The company applies brain research from the Nobel prise winning researchers.
In other words, telepathy is now possible. If I can learn to control a robot through brain signals then I can also send binary information to others, using the power of thought only.
The concept of a digital twin has been around for a while. It is used to describe a digital representation of a physical thing on some abstraction level that continuously maps to the state of the physical thing. Before the introduction of cyber-physical systems this has been called a model.
Digital twins are widely used nowadays when we interact with physical things such as cars or robots. The beauty of the concept is that you can interact with the digital representation mirroring the behaviour of the actual thing you want to control. Controlling hardware involves embedded programming, adaptors and protocols, and these are abstracted away for you. Examples range from an app for your thermostat to a full representation of a manufacturing plant. And telecom radio sites have digital twins as well.
Looking at my 10-year-old son I start wondering how much physical things the new generation craves about. He’s very much into the digital world as many other kids and, unlikely his sister, does not care much about physical stuff. His room is pretty empty and his most important things are his computer and his phone. He’s very quick at spending his monthly allowance on computer games and in-app purchases. His heroes are digital, and physical things are of no interest.
With my background in computer science I do admire the shift towards the digital, and the appreciation of pure-software products, with the willingness to spend money on those. This is a really good trend from the sustainability perspective as well. But empty rooms are really no fun, and I keep buying my 10-year-old nice pillows, pictures and other decorations.
And now it’s time for me to share my business idea with you – up for grabs, first come first served! What if we could bring that digital things that miss the physical side to life through “physical twins”? Similar to Disney et al selling their soft toys looking like Mermaid or Nemo. But a twin for real, acting in real-time and real-life, along with its digital original. And it should not be so big of a step given the 3D-printing techniques, the cost of motors and modems, and the fact that the digital model has already been designed. Wouldn’t you love your favourite character from SIMS walk around in your house? Do you see the scale if we could right-click and order home physical twins of our favourite game characters? Personally, I would immediately invest in a couple of friendly dinosaurs from Lost Eden.
We are constantly looking for new values that can be created out of old technologies that become more mature and scalable. Processors are powerful enough to give us search results in seconds, telecom networks become fast enough to transmit huge amounts of information almost instantly, small connected devices are deployed in everything, we expect mobile internet to work everywhere, and we expect all this to be for free, or almost for free. What we gladly pay for is the “Jet Black” shade of the phone, fancy filters in the camera application and Siri with a sense of humor. And we can easily forgive these value-adding services when they misbehave – it’s not a big deal if Siri does not get what you want from her. But when we lose coverage it becomes very annoying. Or if your device suddenly freezes and restarts, especially when you are right in the middle of credit card payment process. Can I trust that my microphone is actually muted after I clicked on mute? Can I trust that my information will stay secure?
Apple iOS 10 is Apple’s high score in the number of bugs. And here we’re talking about bugs in that basic functionality, such as battery life or sound, that we do not want to pay any extra for. They have already released several updates to it, still failing to fix the sounds problem on iPhone 6S. Isn’t is unacceptable for a phone that costs $600? There are tons of information on the web about fixes and workarounds for the bugs in iOS 10, and even detailed instructions on how to revert to iOS 9. Some people even turn it to something positive: “Apple are so quick with their bug fixes and updates!” But isn’t it exactly what we paid Apple for from the very beginning – things just worked.
If my device tells me that the sound is on when in fact there is no sound, can I trust it’s muted microphone to actually be muted? Does the fancy outfit matter when you stop trusting someone?