Humans hating their AI teammates could be of concern for researchers designing this technology to one day work with humans on real challenges — like defending from missiles or performing complex surgery. This dynamic, called teaming intelligence, is a next frontier in AI research, and it uses a particular kind of AI called reinforcement learning. A reinforcement learning AI is not told which actions to take, but instead discovers which actions yield the most numerical “reward” by trying out scenarios again and again. It is this technology that has yielded the superhuman chess and Go players. Unlike rule-based algorithms, these AI aren’t programmed to follow “if/then” statements, because the possible outcomes of the human tasks they’re slated to tackle, like driving a car, are far too many to code. “Reinforcement learning is a much more general-purpose way of developing AI. If you can train it to learn how to play the game of chess, that agent won’t necessarily go drive a car. But you can use the same algorithms to train a different agent to drive a car, given the right data” Allen says. “The sky’s the limit in what it could, in theory, do.”
Employees in the defense and intelligence sector are in near-constant contact with each other, sharing information often under challenging circumstances. They move files and documents from low trust environments into networks that hold a nation’s most sensitive data, where a data breach could have a serious impact on national security. Consequently, when it comes to sharing any kind of document, these teams cannot risk threats slipping through the net. Human attackers are now using machines to engineer malware at a pace only imaginable a few years ago. Today, it’s possible to engineer a new piece of malware and to make each version of that file suitably different so that it’s almost impossible for traditional malware protection solutions to identify. In the same way that Facebook or Twitter use algorithms to create a truly unique social feed of information that is tailored to the interests and tastes of a user, bad actors can use similar algorithms to deploy essentially the same underlying threats but packaged in ways that simply evade detection.
Cambridge Quantum’s efforts to expand quantum infrastructure got significant backing earlier this year when Honeywell said it would merge its own quantum computing operations with Cambridge Quantum, to form an independent company to pursue cybersecurity, drug discovery, optimization, material science, and other applications, including AI. Honeywell said it would invest between $270 million – $300 million in the new operation. Cambridge Quantum said it would remain independent, working with various quantum computing players, including IBM. The lambeq work is part of an overall AI project that is the longest-term project among the efforts at Cambridge Quantum, said Ilyas Khan, founder, and CEO of Cambridge Quantum, in an e-mail interview. “We might be pleasantly surprised in terms of timelines, but we believe that NLP is right at the heart of AI more generally and therefore something that will really come to the fore as quantum computers scale,” he said. Khan cited cybersecurity and quantum chemistry as the most advanced application areas in Cambridge Quantum’s estimation.
The amount of work you have is driven by the ability of software to make a meaningful difference in your organization. Take a look at your current queue of work. If your team is like most IT teams there will be a mountain of unmet demand for new applications or additional functionality for existing applications. Thinking that any amount of automation will reduce that demand to zero is like thinking that a faster car will get you to Mars. If low code software starts taking some of your work, there will likely be other projects you can work on. If you handle this right, you can even shuffle some of the painful projects over to the party-goers on the low code bus. ... Secondly, and more fundamentally, there are certain aspects of software engineering that are harder to automate than others - making it unsuitable terrain for the low code party bus to drive across. For example, low code tools make it easy for non-developers to create a table to store data. But they can't do much to help the non-developer structure their tables to best map to the business problem they are trying to solve.
When you sign a contract, you expect both parties to hold their end of the bargain. The same can be true for testing applications. Contract testing is a way to make sure that services can communicate with each other and that the data shared between the services is consistent with a specified set of rules. In this post, I will guide you through using Joi as a library to create API contracts for services consuming an API. ... Before we get started, let me give you some background about contract testing. This kind of testing provides confidence that different services work when they are required to. Imagine that an organization has multiple payment services that utilize an Authentication API. The API logs in users into an application with a username and a password. It then assigns them an access token when the log-in operation is successful. Other services like Loans and Repayments require the Authentication API service once users are logged in. ... Contract tests are designed to monitor the state of an application and notify testers when there is an unexpected result. Contract tests are most effective when they are used by a tool that relies on the stability of other services.
Regulators need to know what the technology is capable of, but they need not know every technical detail just to make good law. “If you can understand clearly what the technology is doing, I think that you can make pretty good judgments about what the fundamental financial activity is and what regulatory box that financial activity can or should fit in,” he told Webster. Strip those technologies down a bit, and they boil down to some basic underpinning concepts that lend themselves to governance. At the core of blockchain and cryptos is database architecture, said Gerety. “It has some neat properties, but nowhere else in the financial services industry do you get regulated differently if you use SAP or Oracle,” he said. To get a sense of how one might approach “newness” in a sector, he offered a concept of a matrix, with axes denoting what the future “feels like” and might actually “be.” Babies will pretty much always “be” and “feel” the same. Not much in the way of technology will change the experience or feelings one will have with birthing and raising a child, despite the newness of, well, becoming a parent.
Let’s start with a data science interview question. Usually, as part of an initial screening round for entry level candidate I like to find an example on their CV of a project that used real life data. Real life data is much nastier than academic and research data. Its chalked full of missing data, mixed (integer and string) data and outliers that make consuming and modeling the information grossly more difficult. Invariably most of the conversation revolves around these real world considerations. How do you handle missing data? Usual answers involve some sort of information replacement strategy like replace them with the average value of the column. Fair and reasonable. How do we deal with malformed or mixed data? Again usually a fair answer involving mapping strings to numbers. Finally what did you do about the large outlier events? Usually the answer is that they ‘removed them’ because you ‘can’t be expected to predict rare events.’ The ultimate justification: it improved the models accuracy. That’s good answer if building a forecast is a game or contest, much worse if you want to use it.
This is noteworthy for a couple of reasons. First, it is a recognition that many banks, along with a slew of other financial institutions, are adopting DLT as a technology enabling better processes. Simply put, financial institutions are moving past the exploratory phase of DLT and are now actually implementing the technology into their operations. Secondly, the OCC is declaring its intent to explore and define appropriate governance processes for banks to deploy when such changes are implemented. In other words, the OCC is defining its intent to regulate how such changes should take place. ... The immutability of a distributed ledger provides a new level of security. It is challenging to establish a single customer view across different jurisdictions and business lines. With mutualized data management, DLT allows permitted parties to share data securely and in real time, which could address challenges of Know Your Customer (KYC) and Anti Money Laundering (AML). The themes are clear – DLT injected into the banking and financial ecosystem is an equalizer, a simplifier and a fortifier.
For business intelligence, the airline has been a long-term user of WebFocus from Tibco. It also uses Microsoft PowerBI. Riboulet’s reason for using two BI platforms is because “they complement each other”, each having different functions it finds useful. For example, WebFocus offers Air Canada the ability to push out reports via email, a feature not available in PowerBI. Riboulet says this is useful for people working in operations, who may only have access to their phone and need to see embedded reports. Also, the data team noticed that many business users require similar datasets and attributes, which can be pulled together into pre-built reports. The company also uses the data grid feature in WebFocus to aggregate data in a way that can easily be customised by users and can be exported to Microsoft Excel. It has also deployed WebFocus Hyperstage, as a staging area for data, to avoid direct access to its on-premise database systems. Riboulet views the data team at Air Canada Cargo as internal consultants who discuss data requirements with businesspeople.
If you want to automate your finance function and bring lower costs to operate the financing and accounting needs, taking control can provide you with numerous benefits. This includes prioritization of your processes that align with your strategic vision, controlling resource investments and commitments, and insuring SOX control frameworks are adhered to at the onset. It’s not surprising that some finance organizations can feel underserved by their IT partners, as ITs responsible for supporting the whole organization and finance operations can take a back seat to other priorities. This does not mean that IT should be left aside. IT will have a role, even if you run your own automation program end-to-end, and you will need them to have a seat at the table. You will want to avoid creating a shadow IT group and truly focus your financial resources on process improvement and automation. It’s best practice to leverage your IT team for infrastructure, network security, understanding ERP/system schedules, roadmaps, and disaster recovery processes (at a minimum). It is also recommended to adopt the cloud version of the tools, which can significantly reduce the needs of your IT org
Quote for the day:
"Problem-solving leaders have one thing in common: a faith that there's always a better way." -- Gerald M. Weinberg