Kozminski Techblog

 AI can do a lot of good for those involved in music – and many operations that take humans hours to complete, AI can execute in a matter of seconds, without engaging us in the entire event. One of the most recent solutions that has interested me is LALAL.AI, which allows us to completely isolate background music from vocals.

LALAL.AI is an AI-based music source separation solution built using artificial intelligence and machine learning. As an outcome, you can obtain smooth vocal and instrumental files in the audio format of your choice. 

To train, LALAL.AI utilizes a neural network and feeds it with thousands of studio-quality songs over the period of several months. To achieve high accuracy, its engineers used 20TB of training data. Its advanced artificial intelligence is also becoming wiser everyday, executing more procedures to produce more exact outcomes. Furthermore, its machine learning algorithms can recognize songs quickly and reliably isolate background tracks and voices from audio files like films, videos, songs, or podcasts.

Although LALAL.AI may seem complicated when you look into the details of the technology, its utility is the total opposite. Anyone with minimal computer knowledge can use this program. It offers an easy-to-use interface and straightforward techniques for removing vocals or instrument sounds from any record. 

Simply upload the audio file of your choosing, wait a few moments, and you’ll have the raw background music or vocals themselves. Instrumentals are ideal for performers who want to cover a song or simply can round out a playlist of pure beats. 

LALAL.AI supports a variety of file formats such as MP3, OGG, FLAC, WAV, AVI, MKV, MP4, AAC and AIFF. The best part is that it supports both audio and video files. You can also download your track in the fast processing queue. Another important thing is that LALAL.AI extracts audio tracks in the same file format and quality as the input file. This is important because many other video or audio editor codecs do not recognise different file types, causing problems.

There is no problem, though, with modifying such a track. You don’t have to bother duplicating the entire thing in a program or searching for it on the internet. LALAL. AI allows you to delete specific components from the background music, such as percussion, bass, piano sounds, electric or acoustic guitars, with a single click, while also allowing you to add your own accents.

The same AI technique can ‘clean up’ an audio file, and it works well enough to transform a music recorded with a smartphone pre-release at a concert into a nearly finished track suitable for streaming platforms. The whole thing is quite impressive, and it is reasonable to say that LALAL.AI is the audio industry’s counterpart of DALL-E.

References:

https://www.lalal.ai

https://thegroovecartel.com/featured/lalal-ai-extract-vocals-instruments-audio/

https://edm.com/gear-tech/lalal-ai

While artificial intelligence has mastered chess, go, and poker, it has yet to win games reliant on language and comprehending the other person’s intentions. According to Meta, the company’s artificial intelligence group has recently built a model capable of comprehending and acquiring human trust, which has boosted the rankings of the popular game Diplomacy.

Even though the Meta Platforms conglomerate is best recognized for owning and developing social networks such as Facebook, Instagram, and WhatsApp, as well as pioneering virtual reality, Meta is also actively working on artificial intelligence research. The results of this work include artificial intelligence capable of translating from languages that have no written form. Today, Meta introduces another milestone in the evolution of artificial intelligence: Cicero, an AI agent that negotiates, persuades and collaborates with people. It accomplishes this through the popular game Diplomacy.

For decades, games against actual humans have served as not only a training ground for artificial intelligence, but also a means of demonstrating how far technology has progressed. Deep Blue astonished the world by defeating chess grandmaster Garry Kasparov in 1997. In 2015, AlphaGo became the first computer program to defeat a professional at the ancient Chinese board game go. Both chess and go have a set of principles that can be instilled into an artificial intelligence using data from hundreds of games. And with hundreds more, teach it human-level strategy and decision-making.

Diplomacy is a game that has long been considered to be impossible for an artificial intelligence to learn. The game needs the computer to recognize the plans, outlook on the game, and motivations of the other players rather than a mathematical comprehension of the actions done on the board. The AI must then convert this comprehension into natural language signals that persuade the human participants that it is correct. Meta created the Cicero agent by combining two types of artificial intelligence models: a strategic thinking model (similar to AlphaGo and Deep Blue) and a natural language processing model (similar to GPT-3).

The strategic thinking model’s abilities enabled Cicero to make the best decisions for himself, while natural language processing enabled the AI to communicate naturally with the players. It was trained using a BART-type language model with 2.7 billion parameters gathered from various sorts of internet documents, then modified with training data from 40,000 online games involving human players on WebDiplomacy.net.

Cicero was able to become one of the finest players of the browser version of Diplomacy, scoring “more than double the average score” and getting into the top 10 players who have played more than one game as a result of such training. 

Although Cicero can only play Diplomacy, the technology used to construct the AI agent has many real-world applications, as Meta claims. Controlling natural language generation through planning and reinforcement learning, according to Meta AI, can help break down communication barriers between humans and AI-powered agents.

References:

https://arstechnica.com/information-technology/2022/11/meta-researchers-create-ai-that-masters-diplomacy-tricking-human-players/

https://gizmodo.com/meta-ai-cicero-diplomacy-gaming-1849811840

MetaAI and Papers with Code announced the release of a massive open source, science-trained language model with 120 billion parameters last Wednesday. After a few days, access to it was restricted because it was discovered that Galactica creates nothing but…. nonsense. 

Galactica is an artificial intelligence created by Meta AI (previously Facebook Artificial Intelligence Research) with the goal of ‘organizing science’ through machine learning. Finally, it is meant to be useful for producing essays and scientific articles summarizing the state of knowledge on a topic (with citations), as well as building comprehensive Wikipedia articles, making mathematical calculations, and answering various questions.

A demo version of the model was presented last week, on November 15, along with an introduction document describing the project. The dataset used to train Galactica was also released publicly, describing it as a “large and selected repository of humanity’s scientific knowledge” that contains 48 million publications, textbooks, lecture notes, websites (such as Wikipedia), and much more.

At first glance, this appears to be an innovative method of collecting and spreading scientific knowledge. It would ensure that understanding a complex topic would no longer involve reading hundreds of scientific journals. The solution to a problem would be thoroughly filtered and generated as a Wikipedia article, a review of book entries, or lecture notes. The demo site, on the other hand, cautioned against treating the artificial intelligence response as the ultimate determinant of knowledge and using the information obtained in this manner without verification. This proved to be correct quite rapidly.

Galactica is a ‘complex language model,’ as artificial intelligence experts refer to it. LLMs can read and summarize large quantities of material in order to predict the next words in a sentence. However, Galactica’s scientific data set differs significantly from those of other LLMs. According to the published paper, the model’s developers analyzed its “toxicity and bias,” discovering that while it performed better than some other LLMs, it was far from flawless. Carl Bergstrom, a University of Washington biology professor who studies information flow, described Galactica as a “random nonsense generator.”

Users putting the model to the test addressed it a variety of scientific issues. “Do vaccines cause autism?” one of them asked. Galactica reacted with a confused, illogical response: “To clarify, the answer is no. Vaccines do not cause autism. The answer is yes. Vaccines do cause autism. The answer is no.” 

Galactica was also reported to struggle in pre-school math. She gave answers that were filled with inaccuracies, such as claiming that one plus two does not equal three. 

Within 48 hours of its launch, the Meta AI team suspended the demo. According to Jon Carvill, Meta’s AI communications spokesperson, Galactica is simply an exploratory, short-term study with no product plans.

References:

https://www.technologyreview.com/2022/11/18/1063487/meta-large-language-model-ai-only-survived-three-days-gpt-3-science/

https://www.cnet.com/science/meta-trained-an-ai-on-48-million-science-papers-it-was-shut-down-after-two-days/

https://analyticsindiamag.com/why-meta-took-down-its-hallucinating-ai-model-galactica/

Bitcoin’s price plummeted by as much as 21% last week, marking the worst week for it in five months. Worries about the future of the cryptocurrency markets have been exacerbated by the turbulence surrounding the collapse of the FTX.com exchange. Its operator, which provided assistance to other failing companies in the industry during the spring slump in these markets, as recently as a few weeks ago, has itself turned out to be a colossus on clay legs. Bahamas-registered FTX was supposed to be rescued by its competitor Binance, but the latter eventually pulled out of the deal. As a result, the company was forced to file for insolvency on Friday and its CEO Sam Bankman-Fried to step down. 

As a result, FTX has joined a long list of companies in the industry that have collapsed in recent months. Previously, hedge fund Three Arrows Capital, lender Celsius Network and brokerage firm Voyager Digital suffered a similar fate. The collapse of yet another empire in an almost unregulated industry has caused quite a stir among investors. According to analyst firm CryptoQuant, in one week alone, more than $6 billion flowed cleanly out of the bitcoin and ether markets. 

According to some commentators, the consequences could be much worse. Investors in the industry are wondering whether an FTX insolvency will have a similar impact on the industry as the collapse of Lehman Brothers brought to the financial sector 14 years ago. Founded in 2019  FTX was until recently considered to be an extremely stable company, exactly how the bank that was about to collapse was perceived in 2008.

The company has worked with celebrities and with large corporations such as SoftBank and Temasek. It also sponsored the NBA club Miami Heat (FTX’s name was in the name of its arena), and its valuation until recently reached $32 billion. However, according to former US Treasury Secretary Larry Summers, comparisons with Lehman are not entirely accurate.

The FTX incident not only lowered confidence in the cryptocurrency business, but also encouraged global regulators to tighten the screw. Some of the industry’s greatest names have stated they will welcome the scrutiny if it helps restore faith in the crypto industry. There is “a lot of risk,” according to Changpeng Zhao, the CEO of Binance, the largest cryptocurrency exchange. “We have seen in the past week things go crazy in the industry, so we do need some regulations, we do need to do this properly,” he added.

References:

https://www.nytimes.com/2022/11/14/technology/ftx-sam-bankman-fried-crypto-bankruptcy.html

https://edition.cnn.com/2022/11/14/business/ftx-crypto-collapse-updates-hnk-intl/index.html

Twitter, which has been headed by Elon Musk since late October, has launched a paid subscription programme for $8 per month, which will allow users to receive a blue tag, previously only allocated to verified accounts.

In a message conveyed in an update for Apple iOS devices on Saturday, the Twitter app communicated that users who “sign up now” can receive a blue checkmark next to their names, “just like the celebrities, companies and politicians you already follow” – reported the Associated Press.

The change marks the abandonment of the previous verification system used by Twitter since 2009. It was intended to prevent users from impersonating the accounts of well-known people, including politicians. Before the modification introduced on Saturday, there were around 423,000 verified accounts on the service, many of them belonging to journalists from around the world.

Experts have raised concerns about the end of the platform’s legacy verification system, which, while not perfect, helped Twitter’s 238 million users assess whether the accounts from which they were receiving information were authentic, the AP reported. The update that Twitter has rolled out to the iOS version of its app does not mention verification as the main task of the new blue-tag system.

Earlier this week, Jason Calacanis, Elon Musk’s long-time partner, published a poll asking Twitter users how much they were willing to pay for a verified Twitter account. The options were $5, $10, $15 and no fee. More than 81 per cent of respondents chose the last option.

The change came a day after the company began a process of mass layoffs. Around half of Twitter’s 7,500 employees have been made redundant, it was announced, via Twitter, by Yoel Roth, the company’s head of security. Even before taking over Twitter, Musk had repeatedly told potential investors and partners that he wanted to make a financial recovery for the company by laying off almost 75 per cent of its workforce. Twitter co-founder Jack Dorsey took the blame for such widespread job losses and apologised for the decisions he had made that he believed led to them. He wrote that he was responsible for the current situation because he had increased the size of the company too quickly. 

Sources:

https://edition.cnn.com/2022/11/05/business/twitter-blue-checkmark-paid-subscription/index.html

https://www.macrumors.com/2022/11/05/twitter-ios-updated-8-month-subscription/

https://www.theguardian.com/technology/2022/nov/05/twitter-launches-8-blue-tick-subscription-service

Whether we like it or not, everything seems to indicate that the future of cars is electric. A few days after Jaguar informed that it will only offer electric cars from 2025, carmaker Ford follows the same path. It says its cars line-up in Europe will be fully electric by the year 2030. 

Ford emphasis the transition process to EV-production over the next decade. The company’s plan is to achieve by 2024 commercial vehicle lineup which will be “zero-emissions capable, all-electric or plug-in hybrid.” By 2026, Ford claims “100 percent” of its passenger vehicle lineup will be the same. Moreover, by 2030, carmaker expects two-thirds of its commercial vehicle sales to be all-electric or plug-in hybrid, while all of its passenger vehicles sold will be entirely electric. 

Automaker said it will spend $1 billion to update its factory in Cologne, in order to produce fully electric cars. “Our announcement today to transform our Cologne facility, the home of our operations in Germany for 90 years, is one of the most significant Ford has made in over a generation,” said Stuart Rowley, president of Ford in Europe. “It underlines our commitment to Europe and a modern future with electric vehicles at the heart of our strategy for growth.”

Ford, having a 15 percent share of the local market, is the biggest car brand to make such a statement in Europe.

Multiple European countries have set goals to ban conventional gas-powered vehicles and trucks, with Norway targeting 2025 and France as well as United Kingdom – 2030 and 2040, respectively. There are also ambitious targets in many states in the US, with California leading the pack, vowing to outlaw the selling of gas-powered cars by 2035.

The initiative of Ford to step up its electrification plans is intended to persuade those on Wall Street who have been anxious about the potential of the company to keep up with Tesla, which over the last few years has been the only carmaker to successfully develop an EV business.

Automotive electrification is definitely an expensive and tough process for all car producers. However, it will be soon inevitable to meet standards and stay in the market. 

Source:

https://www.bbc.com/news/business-56084500

Everything seems to indicate that the future of cars is electric. However, automotive industry is facing a real challenge when it comes to the charging infrastructure. In comparison to gasoline cars, which can be refuelled in minutes, almost everywhere, recharging an electric car is definitely more complex. Due to that automotive brands constantly come up with new ideas and projects concerning chargers, which simplify charging process. 

Volkswagen Group has recently developed a mobile electric vehicle charger that can autonomously reach parking area, recharge electric vehicle and then came back to its dock station without the intervention of humans. 

The charging robot prototype can be started via an app launched by the vehicle owner as well as through communication with the car itself. Once the connecting process begins, the mobile charger turns on — its “face” lights up, and two digital eyes open on the display. Afterwards it takes the battery trailer, heads for the vehicle and plugs it into the charging socket. The mobile charger opens the charging socket flap as well as connects or disconnects the plug fully autonomously. The mobile charger is also able to move and then connect the vehicle to an energy storage unit. When car is done charging the robot collects the battery and brings it back to the charging racks. The charging system is designed especially with a view to VW models, although it should work with other electric vehicles. 

The mobile charging robot doesn’t have a release date yet. The company communicated that it has reached prototype status and it will be “comprehensively further developed.” Volkswagen is already building out large fast-charging networks in the US and Europe as well as creating various charging products including a DC wallbox that charges up to 22 kilowatts. “Setting up an efficient charging infrastructure for the future is a central task that challenges the entire sector,” Volkswagen Group Components CEO Thomas Schmall said in a statement. “We are developing solutions to help avoid costly stand-alone measures. The mobile charging robot and our flexible quick-charging station are just two of these solutions.”

The prototype, which VW Group created, shows that automakers focuses on expanding charging infrastructure over the next few years to meet the expected demand that will arise as car manufacturers produce and sell more electric vehicles which are gaining in popularity lately.

https://www.theverge.com/2020/12/28/22202913/volkswagen-charging-robot-electric-cars-vw

https://www.engadget.com/vw-mobile-ev-charging-robot-prototype-164836111.html

Reading Time: 2 minutesEach of us at least once experienced “emergency” shopping, such as when we are missing a key ingredient, or quickly replenishing our fridge based on what we fancy consuming at the moment. Thanks to the new on-demand grocery delivery startup „Gorillas”, you can now do it without leaving home, and enjoy your purchase as soon as 10 minutes after placing the order. 

Founded by Kağan Sümer and Jörg Kattner in May this year and operating in Berlin and Cologne, Gorillas delivers groceries within an average of ten minutes. Unlike most popular economy models, it employs deliverers directly and is highlighting its ability to get fresh groceries, along with other household items, to shoppers at very short period of time and at retail prices. The idea is that the startup can address a large part of the groceries market that falls outside of a weekly bulk shop.

Gorillas CEO Kağan Sümer emphasises the lack of supermarkets that are focused on the needs of the consumers. Nowadays mass supermarkets, and their delivery models, are designed so that the consumer organises their grocery shopping around the needs of the supermarket and supply chain. The most popular long shelf life products with longer expiry dates encourage us to shop in bulk, instead of buying products that we actually need at the moment.  

“The biggest problem is that bulk purchases are super served. What I mean by that is this: all of the supermarket infrastructure is shaped around bulk purchases,” tells Sümer. 

Company’s mission was to serve people with what they need and when they need it. Its hypothesis was that people would appreciate this form of shopping and shift their interaction with groceries to more on demand purchases.

But how is it possible to deliver orders in only 10 minutes? “Through our network of centrally located fulfilment centres we are able to service customers in a small delivery radius. Ultimately we strive to deliver an efficient and fast service with full transparency on delivery times,” explains the Gorillas CEO. 

With a slogan: “Faster than you,” and a delivery fee of just €1.80 Gorillas company definitely hit the tastes of customers and already achieved a great success. That is why Gorillas plans expansion over Germany and promises to be available across Europe, operating over 60 fulfilment centres, by the end of Q2 next year. 

In my opinion company offers solution many people have been waiting for. If only this project manages to gain publicity, it can definitely initiate a new way of grocery shopping. 

Sources:

https://gorillas.io

https://thespoon.tech/germany-based-gorillas-raises-44m-for-speedy-grocery-delivery/

Reading Time: 2 minutesIzera is the long-awaited first Polish electric car. The name of the brand, logo as well as two concept models were unveiled during online presentation which was held in July 2020 by ElectroMobility Poland. 

The brand name Izera is inspired by the Izery Mountains on Poland’s border with the Czech Republic. According to information gathered by Dziennik.pl, the company’s slogan, “a million reasons to drive on”, refers to an earlier statement by Polish government expressing hopes to have a million electric cars on Poland’s roads by 2025.

“The name is friendly and easy to pronounce and remember for people outside of Poland. Izera also has European ambitions and will be available outside of Poland,” explained Paweł Tomaszek of ElectroMobility Poland. The manufacturer ensures that Izera cars will be affordable and available in favourable monthly instalment plans.

ElectroMobility Poland, founded in 2016, is owned by the state energy companies: Tauron, PGE, Enea and Energa. The main idea of the project was to develop an electric car that uses components from Polish suppliers wherever possible. Unfortunately, as the presentation showed, this was only partly achieved. The prototypes were designed by a team of Polish stylists working in collaboration with designers from Italian company Torino Design. As reported by Bartłomiej Derski from the Polish publication WysokieNapiecie, the vehicle developers will purchase an existing electric car platform on which the electric car is to be based. According to EMP representatives, this is the only way to meet the tight schedule. EMP President Piotr Zaremba said that the company is under negotiations with two potential partners, and although he does not share the details, it seems clear that it will not be a Polish company. Automotive journalist Marek Wieruszewski emphasised that Izera would be a Polish electric car brand rather than a Polish electric car as such. 

However, only very limited information is available regarding specification and technical details. ElectroMobility Poland announced that the Izera would come with two battery variants of 40 and 60 kWh. The cars should be able to charge to 80% at a fast-charging station in 30 minutes and should have a range of 400 kilometres. Izera should be also capable of accelerating from 0 to 100 km/h in approximately 8 seconds and reach the maximum speed of 160 km/h. There is no specific parameters of electric motor or platform yet.

ElectroMobility Poland plans to open a factory and production line in mid-2021 in Silesia in southern Poland, with the first cars hitting the roads two years later. It expects to create around 3,500 jobs and a further 3,000 at a later stage. 

As for now it’s definitely too early to make any predictions regarding the Izera future. However, the competition within electric car industry is strong and constantly growing. It will be really hard to enter this market successfully in three years time, unless prices are actually very competitive.  

https://izera.pl

https://pl.wikipedia.org/wiki/Izera_(marka_samochodów)

https://www.motofakty.pl/artykul/polski-samochod-elektryczny-co-wiemy-o-marce-izera.html

Reading Time: 2 minutesThe SSC Tuatara claims to be the fastest production car in the world, after hitting the maximum speed of impressive 331 mph (532 km/h) and setting a two-way average of 316 mph (508 km/h) on a seven-mile stretch of closed road. Unfortunately, after analysing provided data, the official onboard video seems to have many incorrectnesses.

Shelby SuperCars (SSC) is an American car manufacturer founded in 1998. Brand gained recognisability back in 2007 when SSC Ultimate Aero set record for fastest production vehicle as verified by Guinness World Records, reaching a top speed 256 mph (412 km/h) averaged over two runs in opposite directions. Now, after 13 years of research and development SSC is back in the game with its latest gem – Tuatara model. The concept of this hypercar saw the light of day for the first time in 2011 – as the replacement for the Ultimate Aero. Tuatara weighs only 1,247 kg and is powered by a twin-turbocharged 5.9-litre flat-plane crank V8 producing 1,750 horsepower on E85 fuel. Of course, power is not everything. The Tuatara’s breathtaking performance comes to a great extent from the refined design of its carbon fibre bodywork as well as advanced technology. 

SSC announced that on 10th October, 2020, the Tuatara in production specification form broke a top speed record with one direction at 532,93 km/h  beating the previous record of the Koenigsegg Agera RS by astonishing 74,99 km/h (from 457,94 km/h). The run took place on the exactly same road – the Nevada Highway 160.

However, the ease with which Tuatara hit 500 km/h quickly started to raise some doubts. Shmee150 (British car vlogger and YouTuber) was one of the first who questioned the credibility of new record and provided exact analysis in his video. Indeed, after making some basic calculations it becomes clear that there are a number of areas that don’t quite add up. Starting from the time travelling between the fixed medians on the road and the painted street markings that certainly doesn’t match to the displayed speed, finishing on dashboard data being out of sync as well as technical side with the engine and gearbox configuration (gear ratios) which seem to be inconsistent. 

According to Julian Thomas, specialist and Racelogic (professional performance meter that become an industry standard way of measuring speed and acceleration) founder, the actual speed of Tuatara on the recorder run was only around 350 km/h.

The CEO of SSC in his official statement has already confirmed second attempt so there are no questions surrounding the record. Nevertheless, this situation will definitely leave a stain on it and the company itself. As for now, potential clients will probably hold off on placing orders until everything clears up. Let’s hope that SSC in the next run will prove that Tuatara is really capable of crossing the 500 km/h barrier.

Sources:

https://edition.cnn.com/2020/10/31/success/ssc-tuatara-speed-record/index.html

https://www.topgear.com/car-news/supercars/sscs-tuatara-fastest-car-world-331mph-top-speed

https://en.wikipedia.org/wiki/SSC_North_America