Whilst most research is understandably focused on pushing the boundaries of complexity, the reality is that training and running complex models can have a big impact on the environment. It’s predicted that data centres will represent 15% of global CO2 emissions by 2040, and a 2019 research paper, “Energy considerations for Deep Learning,” found that training a natural language translation model emitted CO2 levels equivalent to four family cars over their lifetime. Clearly, the more training, the more CO2 is released. With a greater understanding of environmental impact, organisations are exploring ways to reduce their carbon footprint. Whilst we can now use AI to make data centres more efficient, the world should expect to see more interest in simple models that perform as well as complex ones for solving specific problems. Realistically, why should we use a 10-layer convolutional neural network when a simple bayesian model performs equally well while using significantly less data, training, and compute power? “Model efficiency” will become a byword for environmental AI, as creators focus on building simple, efficient, and usable models that don't cost the earth.
IBM defines a digital twin as follows “A digital twin is a virtual model designed to accurately reflect a physical object”. They go on to describe how the main enabling factors for creating a digital twin are the sensors that gather data and the processing system that inserts the data in some particular format/model into the digital copy of the object. Further, IBM says “Once informed with such data, the virtual model can be used to run simulations, study performance issues and generate possible improvements”. ... So, how do we use our favorite language Python to create a digital twin? Why do we even think it will work? The answer is deceptively simple. Just look at the figure above and then at the one below to see the equivalency between a Digital Twin model and a classic Python object. We can emulate the sensors and data processors with suitable methods/functions, store the gathered data in a database or internal variables, and encapsulate everything into a Python class.
When you have a monolith, you generally only need to talk to one database to decide whether a user is allowed to do something. An authorization policy in a monolith doesn't need to concern itself too much with where to find the data (such as user roles) — you can assume all of it is available, and if any more data needs to be loaded, it can be easily pulled in from the monolith's database. But the problem gets harder with distributed architectures. Perhaps you're splitting your monolith into microservices, or you're developing a new compute-heavy service that needs to check user permissions before it runs jobs. Now, the data that determines who can do what might not be so easy to come by. You need new APIs so that your services can talk to each other about permissions: "Who's an admin on this organization? Who can edit this document? Which documents can they edit?" To make a decision in service A, we need data from service B. How does a developer of service A ask for that data? How does a developer of service B make that data available?
In the payments realm, Mastercard® Healthcare Solutions optimizes the workflow for payers and providers by automating repetitive and error-prone operations, such as billing and claims processing. According to CIO magazine, many hospitals are now using AI to automate mundane tasks, reduce workloads, eliminate errors and speed up the revenue cycle. The author notes AI’s effectiveness for reducing incorrect payments for erroneous billings, and for preventing the labor-intensive process of pulling files, resubmitting to payers and eventual payment negotiations. ... The successful use of AI for FWA prevention is increasing in popularity. A recent study by PMYNTS revealed that approximately 12 percent of the 100 sector executives surveyed use AI in healthcare payments, three times the number using AI in 2019. Nearly three-quarters of the 100 execs plan to implement AI by 2023. ... These are all important factors when building an AI model and show the need to demonstrate return on investment (ROI) through a proof of concept.
“As we surround applications with our capabilities, we will understand the traffic flow and the performance and what’s normal,” Coward says. “The longer you run the AI within the network, the more you know about what typically happens on a Tuesday afternoon in Seattle.” A key aspect of SevOne is the ability to take raw network performance data from sources–such as SNMP traps, logs in Syslog formats, and even packets captured from network taps–combine it in a database, and then generate actionable insights from that blended data. “The uniqueness of SevOne is really that we put it into a time-series database. So we understand for all those different events, how are they captured [and] we can correlate them,” Coward explains “That sounds like an extraordinary simple things to do. When you’re trying to do that at scale across a wide network where you literally have petabytes of data being created, it creates its own challenge.” The insights generated from SevOne can take the form of dashboards that anyone can view to see if there’s a network problem, thereby eliminating the need to call IT.
The rapid deployment of AI into societal decision-making—in areas such as health care recommendations, hiring decisions, and autonomous driving—has catalyzed ongoing ethics discussions regarding trustworthy AI. These considerations are in early stages. Future issues could arise as tech goes beyond AI. Focus is intensifying on the importance of deploying AI-powered systems that benefit society without sparking unintended consequences with respect to bias, fairness, or transparency. Technology is increasingly a focal point in discussions about efforts to deceive using disinformation, misinformation, deepfakes, and other misuses of data to attack or manipulate people. Some tech companies are asking governments to pass regulations clearly outlining responsibilities and standards, and many organizations are cooperating with law enforcement and intelligence agencies to promote vigilance and action. ... Many technology organizations are facing demands from stakeholders to do more than required by law to adopt sustainable measures such as promoting more efficient energy use and supply chains, reducing manufacturing waste, and decreasing water use in semiconductor fabrication.
Everything is connected in some way, well beyond the obvious, which leads to layer upon layer of real world complexity. Complex systems interact with other complex systems to produce additional complex systems of their own, and so goes the universe. This game of complexity goes beyond just recognizing the big picture: where does this big picture fit into the bigger picture, and so on? But this isn't just philosophical. This real world infinite web of complexity is recognized by data scientists. They are interested in knowing as much about relevant interactions, latent or otherwise, as they work through their problems. They look for situation-dependent known knowns, known unknowns, and unknown unknowns, understanding that any given change could have unintended consequences elsewhere. It is the data scientist's job to know as much about their relevant systems as possible, and leverage their curiosity and predictive analytical mindset to account for as much of these systems' operations and interactions as feasible, in order to keep them running smoothly even when being tweaked.
Like any public blockchain, the open-source code is viewable by the public. Since there is no human being in control, users can be certain the code will execute according to the rules it contains. As the industry saying goes, “code is law.” DAOs are controlled by a type of cryptocurrency called governance tokens, and these give token holders a vote on the project. The investment is based on the idea that as the platform attracts more users and the funds are deposited into its lending pools, the total value locked (TVL) increases and the more valuable its tokens will become. Aave has nearly $14 billion TVL, but the AAVE token is not loaned out. The Aave protocol’s voters have allowed lenders to lock 30 different cryptocurrencies, each of which has interest rates for lenders and borrowers set by the smart contract rules. Different protocols have different voting rules, but almost all come down to this: Token holders can propose a rule change. If it gets enough support, a vote is scheduled; if enough voters support it, the proposal passes, the code is updated, and the protocol’s rules are updated.
It is well understood that blockchain-based digital identity management is robust and encrypted to ensure security and ease of portability. Hence, mandating its effective incorporation for improving the socio-economic well-being of the users, which is mainly associated with digital identity. With time and advanced technologies, digital identity has become an essential entity that enables users to have various rights and privileges. Although Blockchain has various benefits while managing digital identities, it cannot be considered a panacea. Blockchain technology is continuously developing, and though it offers multiple benefits, there also exist various challenges when aiming to completely replace the traditional identity management methods with the latter. Some of the known challenges include the constantly developing technology and the lack of standardization of data exchange. Considering the benefits that come with transparency and the trust earned through blockchain frameworks, numerous organizations are merging to ensure interoperability across their borders.
Data lakes will continue their dominance as essential for enabling analytics and data visibility; 2022 will see rapid expansion of a thriving ecosystem around data lakes, driven by enterprises seeking greater data integration. As organizations work out how to introduce data from third-party systems and real-time transactional production workloads into their data lakes, technologies such as Apache Kafka and Pulsar will take on those workloads and grow in adoption. Beyond introducing data to enable BI reporting and analytics, technologies such as Debezium and Kafka Connect will also enable data lake connectivity, powering services that require active data awareness. Expect that approaches leveraging an enterprise message bus will become increasingly common as well. Organizations in a position to benefit from the rise of integration solutions should certainly move on these opportunities in 2022. Related to this trend (and to Trend #1 as well): the emerging concept of a data mesh will really come into its own in 2022.
Quote for the day:
"The greatest leader is not necessarily the one who does the greatest things. He is the one that gets the people to do the greatest things." -- Ronald Reagan