Kozminski Techblog

Layoffs – the word that sends any Big Tech worker into a nervous spiral, the greatest enemy of middle managers and “a day in my life at Google” influencers, and simultaneously the supposed answer to any business problem, at least according to most consultant firms (we’re not naming any names, but McKinsey, I’m talking about you). 

In the first months of 2023 over 130k people have been laid off from tech companies, following 160k+ laid off over the course of 2022, according to layoffs.fyi – a service made specifically to track how many positions have been cut since the Covid-19 pandemic hit. These numbers are objectively alarming, being compared to the Great Depression and the 2008 economic crisis, and are leading to rising long-term unemployment in the US. It is also worrying how many people cannot find jobs after being laid off, even with years of experience working for tech giants. While some news outlets report that most workers are comfortable enough after being laid off and “show little anger towards their employer”, there are stories of Google employees receiving the dreaded “you’re fired” email at 6 am, getting company resources cut off while travelling for work, and lining up at office security to check who’s fired by seeing if they can get into the building – kind of hard to believe that everyone loves layoffs and nobody’s offended, right?

The reasoning behind each layoff we see is very similar: economic downturn, macroeconomic factors, aftermath of Covid, and sometimes, overhiring. At this point, the last reason seems the most probable – many companies are reporting record profits, so why the rush to save money? 

It appears the actual understanding of some C-levels is that they’ve hired way too many people to do “fake work”. Many tech giants did hire talent just to keep competitors from getting to them first, without giving people actual valuable roles, leading to heaps of middle managers or Operational Directors of Nothing. It’s become a running joke in the industry over the years, and it’s especially supported by stories of employees who talk about writing two emails a day or even having to fight to find work to do after being hired. A lot of people’s jobs were mostly just going to meetings, being in the office a couple of days a week, and enjoying the benefits of working in tech.

However, recognizing that this phenomenon does exist, mostly by the companies’ fault, it’s important to not take the concept of fake work too far – a lot of the time anything that doesn’t have a direct result, like a complete product, or isn’t tangible labor can be interpreted as not work at all, especially by supervisors who want to see something concrete and feasible all the time. Corporate work, on many levels and in a lot of industries, creates itself and does not provide much real value. There are always Project Managers for small long-term projects, additional designers that make sure every part of the product is perfect, administrators and assistants that help with the workflow, and some employees that just make day-to-day tasks more manageable. But if we’re removing all the layers of organization that have been built for years without a plan as to what will replace them, we might be looking at long-term chaos and employee dissatisfaction – guess we’ll see how that works out for Big Tech.

Are they always simpler and better?

Today, many people notice in their checks (bills) already fixed tips, which can be paid directly through specified payment platforms and touch screens.

At the same time, many consumers do not pay attention to the collected funds. It’s even grown to the point of federal and government crackdowns, and companies and online payment platforms (Square and Toast among the most famous) have started deceiving their customers through convoluted technical developments.

From another point of view, namely the developers of such technologies, there are positive aspects to this, because the design of the programs is created in such a way that direct consumers, regardless of a certain pressure, still press that button.

Researchers believe that an important aspect is the very principle of this UX & UI design, which is specially designed to influence the behavior of consumers and their least resistance.

Modern payment technologies have a selection of options for further “donation” of tips, these are certain sets, and maybe even algorithms developed by experts based on a database, which in a certain growth offer percentage deduction depending on the client’s wishes, for example, 10, 15, 20 percent. But there are additional features like “no tip/don’t leave a tip this time” and “choose your amount”.

In a way, the small fixed buttons with the percentage amount in advance seem to force the consumer to choose a higher amount and leave a rather generous tip in the end. This is the success of the modernization of such companies, with the help of the so-called HumanAI.

Reference list: https://www.nytimes.com/2023/03/01/technology/personaltech/tipping-defaults-digital-payments.html

Recently, one of the most widespread topics has become the possibility of artificial intelligence. Engineers are working more and more on a chatbot for Google.

One worker pointed out that AI systems only mimic real conversation – at least for now. It is quite possible that such a program will be able to be fully updated to meet the needs and stated plans, said Neil Sahota, the UN’s leading adviser on artificial intelligence.

Why don’t chatbots currently have feelings or emotions?

After all, many of them are filled with a huge amount of information (codes, databases, and other Internet resources)…Chatbots are created based on the principle of analyzing individual patterns, which in turn are based on the “obvious” reactions of people to situations already known to them.

The big drawback is that even with these capabilities, chatbots are programmed to simply follow human instructions. By themselves, these algorithmic programs cannot improve and learn certain emotions. “So you can’t have a chatbot that says, ‘Hey, I’m learning to drive,’ that’s general artificial intelligence (albeit more advanced) that doesn’t exist yet,” Sahota says.

Engineers from the famous social platform – Facebook, back in 2017, conducted some research on the conversation of one chatbot with another and found that two chatbots cannot normally respond to words, not to mention emotions. Later, the most popular chatbots in the world, Bob and Alice were taught to negotiate about such objects as hats and balls, and in the absence of human intervention, at least on such topics, they did manage to talk.

So, according to many researchers, the most likely path to algorithms with feelings is to program them to want to improve, rather than just teach them to identify patterns and recognize only a fairly limited range of actions. However, it is still difficult to implement, and it is not known when it will be possible to realistically endow artificial intelligence with human empthy.

Reference list: https://www.bbc.com/future/article/20230224-the-ai-emotions-dreamed-up-by-chatgpt

Today, with the development of the latest technologies, the treatment of serious diseases is becoming less and less problematic, however, it has long taken a very large volume of different samples and an even greater number of errors to determine the correct chemical compounds. As mentioned earlier, the world does not stand still and artificial intelligence could predict the composition of a new drug molecule just like any search query in a browser. OpenAI’s ChatGPT helps generate human-like responses for the analysis and synthesis of proteins, the building blocks of life and many medicines.

Will AI be able to correctly apply certain natural language algorithms that quickly analyze speech and predict the next step that can be applied to this biological data? 

“Technologies like this will begin to cover areas of biology that are ‘untreatable,'” says Sean McClain, founder, and CEO of Absci Corp., a drug development company in Vancouver, Washington.

So if these algorithms work as hoped, they will bring new power to AI’s promise to transform drug discovery. There have already been previous attempts, but unfortunately, their use at the time faced limitations of technology or lack of data.

Since the technology is still in its early stages, most medical companies are more focused on using protein language models to improve known molecules. Absci and Helixon are also working with drugmakers to develop cancer drugs (companies positioning themselves as next-generation AI therapeutic discoveries)

 A Merck representative said the company has entered into several collaborations to explore the potential of artificial intelligence in drug development. Helixon signed contracts with two major pharmaceutical companies last month to fight previously untreatable diseases, CEO says 

Reference list: www.bizjournals.com

Companies in the Polish market continue to search for and develop new information and Internet platforms, in particular those that help sales and buyers of products. One of the leading positions in this market is occupied by the company IdoSell.

This online platform uses many options to expand marketplaces and tie-ups with side courier companies around the world. It’s no secret that for the sake of more sales, stores adjust the personalization of the product, focusing on the target audience.

Another interesting trend is omnichannel. Consumers are used to freely choosing between online and offline stores of the same brand. This broadens the horizons and allows choosing, and as a result, the brand gains more trust.

One of the most convenient aspects is the possibility of instant payments, mobile sales, and social shopping, which allows them to buy products without visiting other websites of certain brands.

The presence of artificial intelligence helps to automatically track the previous search history and find the right products as quickly as possible using voice commands.

According to preliminary forecasts, the use of chatbots will also grow. First, they will develop a level of personalization—they will welcome customers, offer assistance, answer common questions, solve customer problems, and deliver products that meet the immediate needs of consumers. To get the most complete information about the product without visiting an offline store, augmented reality will help. By the way, VR is already being used, especially by IKEA. The use of drones for delivery is so far mastered by big brands, namely Amazon, FedEx, and Alibaba, but only in some parts of the world.

Reference list: The research was carried out by Gemius

A team of researchers from the University of Liverpool has recently made strides in the field of molecular electronics, which focuses on the use of molecules as building blocks for electronic devices, potentially paving the way for smaller, faster, and more energy-efficient electronics in the future. The researchers developed a new technique for precisely positioning molecules on surfaces, which is crucial for creating functional electronic devices using individual molecules as components. Their work has the potential to revolutionize the field of electronics and contribute to the development of innovative technologies that could transform industries such as healthcare, transportation, and communication.

This breakthrough could also lead to the creation of more sustainable electronics, as smaller and more energy-efficient devices consume fewer resources during production and use. The team’s findings were published in the journal Nature Communications, and their research could potentially lead to new technologies and partnerships with industry leaders in the electronics sector. Their work also opens up new possibilities for the development of molecular-scale sensors, which could have applications in fields such as environmental monitoring, food safety testing, and medical diagnostics. The team’s innovative approach to molecular electronics could also lead to advancements in quantum computing and information processing, which rely on the unique properties of individual molecules to perform complex calculations and store data. The team’s research could also have implications for the development of more efficient and sustainable energy storage technologies, as individual molecules could potentially be used to store and convert energy at a molecular scale.

Overall, the University of Liverpool’s research in molecular electronics has the potential to revolutionize multiple industries and pave the way for a more sustainable and technologically advanced future. It will be exciting to see how this breakthrough can impact the future of electronics and related industries, as researchers and industry leaders continue to build on this research and explore its many possible applications. As this technology advances and becomes more widely available, it could also lead to new opportunities for entrepreneurship and innovation, as new companies emerge to explore the many possible applications of molecular electronics.

The United States is preparing to ban exports to Huawei in what is expected to be the first in a series of new restrictions that will affect China’s semiconductor industry, artificial intelligence and quantum computing, which are critical to military and commercial technological superiority. fields. sectors. According to White House insiders quoted by the Financial Times, the Joe Biden administration has suspended licensing US companies to export to Huawei and plans to completely ban the sale of US technology to the Chinese telecommunications giant. This is expected to create additional growth challenges for major Chinese chipmakers such as SMIC and YMTC, as well as smaller local competitors such as Hua Hong Semiconductor. End-use control is intended to limit China’s technological development to a fixed technological level, in line with previous comments by US National Security Adviser Jake Sullivan that the US would no longer use a “sliding scale” to maintain its leading position in various sectors.

The White House is also considering a plan to ban US companies from entire segments of China’s tech industry, Politico reported Jan. 27. Sources cited by Politico. Paul Triolo, China tech expert at Albright Stonebridge business consultancy in Washington, D.C., commented that the new end-use controls are designed to automatically detect other Chinese companies like Hua Hong that want to upgrade their technology to work with 14 -nm process technology. Triolo believes the whole system is set up to automatically keep China at a fixed technological level.

EE Times interviewed Keith Krach, a former under secretary of commerce during the Trump administration and one of the drafters of the Chip and Science Act, to get his point of view. Kra believes that SMIC and YMTC, among dozens of Chinese companies, pose a serious threat to US national security because they produce semiconductors for their military and, as a result, should be subject to additional restrictions. The collapse also suggests giving priority to Alibaba, Baidu and Tencent, as well as their subsidiaries, as they are the most important companies in China’s artificial intelligence military program and second only to Huawei in securing their state of surveillance.

Israeli chip startup Hailo has launched a new series of artificial vision chips called Hailo-15, which are designed for different use cases than its flagship Hailo-8 chip. While the Hailo-8 was primarily used as an image processing co-processor in industrial peripherals and aggregators, the Hailo-15 series is a system on a chip (SoC) designed for IP cameras. Hailo-15 includes an integrated accelerator, CPU and DSP to provide heterogeneous computing capabilities. The Hailo-15 series provides less processing power than the Hailo-8, and its AI accelerator architecture is “similar to Hailo-8 but improved.”

The Hailo-15 series is designed for multitasking, making it ideal for multi-sensor smart cameras. It offers 7 to 20 TOPS of INT8 performance, depending on the part number. Hailo-15H offers the highest performance with 20 TOPS, followed by Hailo-15M with 11 TOPS and Hailo-15L with 7 TOPS. The power consumption of the chips depends on the load, but today customers use less than 2 watts.

Hailo CEO Orr Danon said customers are looking for solutions that will allow them to reuse their software investment across multiple environments, whether it’s a camera or a gateway. The Hailo-15 series solves this problem by using the same technology stack as the Hailo-8, which is new and improved, yet fully compatible with various forms of deployment.

Customers usually buy the Hailo-8 or Hailo-15 series, but some installations may use both. For example, image processing can be performed at the endpoint, and streams from multiple cameras are combined at the gateway. The Hailo-15 series offers customers the flexibility to use third party CPUs and DSPs without a RISC-V license. Danon stated that DSP is an effective tool to bridge the gap between CPU and AI accelerator by offering a more efficient AI processing solution.

In addition to improving image quality, AI can be used for multiple tasks at the same time, including low-light performance and digital zoom. These applications are common in smartphone cameras but not yet implemented in IP cameras. Hailo’s new product line offers a cost-effective and compact solution for customers looking to expand the capabilities of their IP cameras.

The National Cybersecurity Strategy, unveiled by the Biden-Harris administration, aims to shift responsibility for the security of IoT devices and software from end users to manufacturers and developers. The 39-page paper highlights that too much of the responsibility for cybersecurity falls on the end user, and aims to balance that responsibility so that it is more efficient and fair. The strategy recognizes that software and systems are becoming increasingly complex, benefiting businesses and consumers, but also increasing our collective insecurity.

The document identifies five pillars around which the strategy is built. The first component aims to protect critical infrastructure to give the American people confidence in the availability and resilience of critical infrastructure and the essential services it provides. The second pillar aims to undermine and eliminate threat actors using all the tools of national power to render malicious cyber actors unable to threaten the national security or public safety of the United States.

The third pillar aims to create market forces for security and resilience by holding those within the digital ecosystem best placed to mitigate risks and address the impact of poor cybersecurity on the most vulnerable. This will make digital ecosystems more reliable.

The fourth pillar is about investing in a sustainable future through strategic investment and coordinated joint action. The United States will continue to lead the world in safe and sustainable next-generation technology and infrastructure innovation. This includes prioritizing research and development in areas such as post-quantum cryptography, digital identity, and clean energy infrastructure, as well as developing a diverse national cybersecurity workforce.

Finally, the fifth component aims to establish international partnerships to achieve common goals. The United States strives for a world where responsible state behavior in cyberspace is expected and encouraged, and where irresponsible behavior is isolating and costly.

The strategy also notes that too often new features and technologies are superimposed on already complex and fragile systems at the expense of security and resiliency. The widespread adoption of artificial intelligence systems, which can act in unexpected ways, increases the complexity and risk associated with many of our most important technological systems.

Overall, the National Cybersecurity Strategy released by the Biden-Harris administration aims to address the growing complexity and risks associated with cybersecurity. By shifting responsibility to manufacturers and developers, investing in a sustainable future, and forging international partnerships, the United States is committed to making cyberspace more secure and trustworthy.

Classiq has launched a program dedicated to research and education in the field of quantum software for universities and educational institutions. The program offers access to the advanced Classiq quantum computing platform and Microsoft Azure Quantum cloud access to quantum computing, allowing university professors, students, and researchers to teach courses and conduct research on all aspects of quantum computing.

Classiq CEO Nir Minerbi said the program is an integral part of the company’s broader strategy to expand the reach of the platform and grow the quantum computing business. The program will provide students with the tools and knowledge they need to learn the practical skills of quantum software development, as well as provide researchers with optimized tools for developing advanced quantum computing algorithms.

Minerbi believes the program will have a significant impact on the quantum computing community by promoting education and research in the field and helping to drive innovation and progress in the industry. Classiq and Microsoft are among the leading companies developing software for quantum computing. The quantum stack developed by the firms advances Microsoft’s vision of quantum programming languages.

The Classiq platform is designed to model algorithms at a much higher level of abstraction than the low-level tools that have been used so far, improving design and implementation methods. Minerbi notes that the future of quantum computing is being driven by exciting developments in hardware, with computing power and qubits scaling faster than previously thought. However, there are significant challenges when it comes to programming quantum software that can take full advantage of this growing power.

As a leader in quantum computing software development automation, Classiq is well positioned to drive the growth of the industry. Its platform is designed to be scalable and flexible, with the introduction of high-level user-defined functions, automatic hardware optimization, and intelligent circuit synthesis. With the Classiq platform, users can already develop quantum code for computers with more than 1000 qubits.

The differences between classical and quantum programming make it difficult to find quantum programmers due to the difficulty of creating quantum software. Quantum programmers must have knowledge of both software development and quantum physics. The Classiq program will allow researchers to test, validate and run their quantum programs on real hardware, providing valuable real-world experience. Ultimately, the program will have a significant impact on the quantum computing community, promoting education and research in the field, and spurring innovation and progress in the industry.