Kozminski Techblog

We all know famous quote “Wake the f*** up samurai, we have a city to burn” or the term of describing something as “breathtaking”. Both of them were said by Keanu Reeves and there is no doubts that marketing strategy for Cyberpunk 2077 was excellent (8 million preorders were sold). However, the game by itself didn’t fulfil players expectations.

An exorbitant number of mistakes completely overwhelm Cyberpunk 2077. Flying cars, immortal opponents or unopenable doors are only a few bugs in the ocean pool of problems. Especially console players took a knock, poor optimalization for PS4 and Xbox One resulted removing game from PS Store and Microsoft Store. While programmers of CD Project RED are working in the sweat of the brow, a board of directors has decided to take a floor. On 13^th of January 2021 Marcin Iwiński – a co-founder of CD Project released a video on official YouTube profile of Cyberpunk 2077 in which he apologize on behalf of the company for the flawed product. As a source of problems he points a poor communication between board members. Moreover, Iwiński calls fans to stop blaming workers for the final stage of product. In the video he releases a roadmap with a patching priorities for 2021 and promises to fix all bugs. From the roadmap we can read that the first patch is coming soon and at midyear we can expect DLC which is supposed to be free as a compensation.

Personally I haven’t played Cyberpunk 2077 yet but I truly believe that it is a wonderful production which has huge potential. I’m a big fan of The Witcher 3 (spend 160h in the game) and I pin one’s faith on the new production of CD Project RED and really wait for the first good opinions about patches.

bibliography:

Nowadays Artificial Intelligence is ubiquitous, from e-learning to ordering a cheeseburger in McDonald. Each of our life aspect is more and more depended form machines. However, machine learning is a long, time-consuming process of manually adding data/data sets into algorithm and training them. But what if IA could learn themselves, rely more on data they collect day by day rather then on data which human being provide ? This kind of AI already exist and there are a few advantages about it.

The first big merit of these is called semi-supervised learning. This is where small amount on labelled data is combined with large amount of unlabelled data, and used to learn IA. It’s like having 20 flavours of ice-creams and saying to AI to name 300 other flavours of ice-cream and categorize them. IA have to thing by itself, try to guess, “taste”. Alexa is an wonderful example of self-learning IA. First of all teaching model was trained on thousands hours of labelled data (in this case speech recordings).Then the model is trained on unlabelled data  to become smart IA which can learn by itself.

The second, very significant step. AI will self-learn based on the users feedback. If AI do not understand your request it may ask about repetition. It detects errors and tries to reformulate request, correcting algorithm at the same time. I’ve already talk with this kind of IA. 5 months ago I called Jeans Louis David (chain of hairdresser salons) to order a visit as I used to it. When I heard the voice of IA in my receiver I was surprised. Nevertheless I didn’t find any difficulties in ordering visit. It took me 3 min to answer all questions which were required, so it shows that IA assistants works and it works pretty well.   

As we can see, IA is more and more independent. In 2021 we will see a growing body of self-learning IA with minimal human intervention. Even though machines still cannot taste ice-creams I think it’s only matter of time and development of IA.

Based on:

The Wired World in 2021 01/2021 “In 2021, AI assistants will teach themselves” by Rohit Prasad

Reading Time: 2 minutesWhen in march 2020 Polish government announced a two weeks lockdown for the whole country people and especially students took it as a short spring break. Nevertheless, no one imagine that time that we will end up working 5-9 hours each day in front of computer screens for over 9 months. It was not only shook for employees who switched from traditional work office  to remote office but also for student who found e-learning as extremely insufficient way of acquiring knowledge. Math teacher who explains you a new topic on whiteboard by recording a lesson in 144p quality or connection issue of your English teacher are only a few students nightmares nowadays. However, 2021 will be a year of artificial intelligence (also in the education field) working arm by arm with teachers who wants to deliver the best lesson as it’s possible.

China leads the mass-scale e-learning solution which combine AI and human teachers. This technological country has 9 unicorns (include VIPKid, Zuoyebang, Yuanfudao)  which allow schools to enforce this hybrid style. Those companies try to help children making progress rather than provide them traditional lessons that leave some students behind and other board.

Imagine that 6-years-old Ling write a story full of complex character which IA visualizes and identifies. It is possible by using new image-recognition software provided by Hexiaoxiang Network Technology – a literacy platform. The program which based on smart machine learning and  database gives child a video with instruction and feedback.

We can find four key areas in education: teaching, learning, practising and testing. There are already companies which improve all of them by developing IA, VIPKid is one of them. They’ve created IA-embedded class which includes animated characters who assists the human teachers. The results of this combination was astonishing. Course-completion went from 80% to 90% and the rate of correct answer increased form 50% to 80%. All of those results generate data which can easily improve program and make it even better.

Thera is no doubts that technology is thriving and improving year by year. Today IA assistant book your hotel, tomorrow programs will tech you how to behave in new workplace. Outbreak of COVID-19 hit unexpectedly and we had to adjust to new, IT environment. Students are the most  aggrieved in this situation but with new education platforms there is still hope for better tomorrow, tomorrow with new and remarkable educational system.

Based on “AI will enhance remote learning” by Kai-Fu Lee – The wired 1/2021

https://schoolsweek..co.uk/coronavirus-schools-have-legal-duty-to-provide-remote-education/

Reading Time: 3 minutesIt is believed that Jaffe Bezos is peculiar god of our times and we shouldn’t be surprised by saying that especially in US where 57% of American households are members of Amazon Prime in comparison of 51% Americans who attend church. The question which came to my mind is – what is more impactful and powerful in our world, Jeff Bezos and his Amazon or religious ? For most young and hungry of success teens who dreams of  becoming entrepreneurs Jeff sounds like oracle  and his new vision of autonomous vehicles seems to be  a prophecy.

Delivering crucial good like groceries, medicines or cosmetics have never been more important than during the outbreak of COVID-19. A vast majority of consumers switch form going physically to the shop to ordering things on-line where they can easily find their precious commodities. This type of behaviour create a kind of a  rate-race in environment of delivering companies where Amazon wants to lead and dominate. In order to sustain constant delivers Amazon bought 100 million of masks and 31 000 thermometers for prize of  $4 billion to create a safe- space for the workers of warehouses and delivers.

• 
• https://www.therobotreport.com/amazon-autonomous-vehicle-invest/

While it’s a bit too late to develop new technologies for current challenge which is facing COVID-19, (especially when we see 2 prototypes of  vaccines on the horizon) CEO of Amazon things about autonomous vehicles in more extend way. He wants to replace people by self- driving vans, small bot and drones buzzing over our heads, not only to contain Amazon’s moves in future epidemic but also to increase an efficiency and speed of delivering goods.

Moreover, the costs of shipping commodities is exorbitant. In 2018 the company spends $27 billion which is 23% more in comparison to previous year. According to McKinsey & Company research Amazon can save over $10 billion per year on switching to autonomous vehicles.

To accomplish the goal of replacing delivers, Amazon already made a few significant steps. In 2016 they acquired a patent for system of autonomous vehicles which enable them to find proper way in suburban sprawl to deliver orders. Furthermore, in partnership with Toyota, Amazon is developing a concept of self- driving vehicle called e-Pallete which would be capable to move people and packages. They also invested around $530 million in Aurora – a Silicon Valley self-driving vehicle star up which develops an IA brain in in autonomous vehicles.

https://www.newsweek.com/how-amazons-bet-autonomous-vehicles-can-help-protect-us-viruses-1504169

While Amazon is still working on their concept, first self- driving delivery already appeared on streets of San Mateo, California. It was created by Silicon Valley start up Udelv for Draeger’s Market. It works quite similar as Uber. You order the package for certain address and vehicle  rolls up to your home. Whereas the van is getting closer to the destination you receive an SMS-s with a code which you later punches into the screen of vehicle. It opens door of van  which contain your delivery. So simple but complicated to create at the same time.

Speaking about delivering goods, let’s don’t forget about air zone and drones which may be very useful in shipping packages. Jeff Bezos said in one interview (CBS’60 Minutes in 2013) that they are thinking about deploying a fleet of drones which would deliver orders to 5 pounds (according to Amazon’s data, 86% of all packages weight under 5 pounds).In China, online retire JD.com has decided to use drones, cutting time of delivering orders is mountain villages  form day to minutes and reducing the costs dramatically.

Thinking about drones we take into consideration a numerous of advantage as speed of deliver or emit less greenhouse gas but we all know that every medal has two sides. A 2017 NASA study found that sound of heavily dense downtown is less annoying then group of drones buzzing over our heads. Moreover, there is many regulations and rules which Amazon has to respect to deploy a fleet of drones.

To sump up, a concept of autonomous vehicles in Amazon is a big deal. It’s a complicated process of buying licences, investing in many star ups, defeating obstacles and developing their own product. So if you want to be a successful entrepreneur, look at the oracle of Bezos and analyse his moves, decisions or pray to the god for good idea for your start up, the choice is yours 😉 .

Source: Newsweek Learning English 4/2020 ” Amazon’s Autonomous Vehicles” by Brian Dumaine

Reading Time: 2 minutesWine is not an easy topi,  finding perfectly suiting bottle to specific occasion or dish sounds like challenging task. Every label conceal many information starting from region, appellation or producer, ending on graves variety and place of bottling. Heini Zachariassen was an amateur not a sommelier and standing in front of wall with wines in supermarket made him headache so he decided to solve that problem and create Vivino.

Vivno is an application where you simply take a picture of label and after few seconds you get a extensive profile of bottle – prize, region it came from, grapes variety of even proposal of pairing with food. I personally use this app to create my own “data base of wines” to check a rating or structure of wine. Moreover, every time I can go back to my recently scanned labels and compare them with the one I’m currently drinking. But  the main question that  come to mind is – how Vivino accomplish such a big data base of over 10 million different labels and 1.3 billion photos in 10 years of existing ?

At the beginning they had no clue how to provide information about every bottle users scan. They played catch-up with what users needed but it was too little to please everyone so they sent to Indie as many labels as it was possible to find. There people were doing research of  labels, digging and looking  for every crucial information to create a basic profile of wine. After 18 months they achieved a 70% of successful scans what encourage people to use it again. From that moment they had noticed a significant uptick in giving rates and reviews in their application, what brings them to develop another aspect of wine profile – taste.

They made an algorithm which is catching a key words from reviews, words which are responsible for smell and taste of wine. The program simply searches names of  fruits, vegetables or flowers and based on that create a taste profile of wine. I think is a magnificent idea, nowadays most of us reading an obligatory chapter of book or an article are looking for those key words. They smoothly bring this concept to Vivino upgrading it to another level.

Today Vivino has more than 50 million downloads and around 42 active users, it’s a major wine applications without any competition on the horizon. Moreover, on 25 March of 2020 Vivino noted biggest sales of wine in their 10 years history ( vivno is also a marketplace where you can purchase wine) beating their previous record from Black Friday 2019. Let’s take in consideration the fact that during Corona pandemic gastronomic applications like uber eats or smaczne.pl undergo renascence and acquire many new users day by day. Vivno belons to the same group, and since wine market do well during pandemic Vivino doesn’t complain.

Bbliography:

https://www.linkedin.com/pulse/how-vivino-built-worlds-largest-wine-marketplace-from-marian-gazdik/

https://www.forbes.com/sites/parmyolson/2016/06/21/vivino-retail-partnerships-the-amazon-of-wine/?sh=45a36efa2251

https://finance.yahoo.com/news/how-the-vivino-app-grew-its-user-base-by-29000-by-helping-people-find-the-perfect-wine-120329338.html