Daily Tech Digest - September 28, 2023

What is artificial general intelligence really about?

AGI is a hypothetical intelligent agent that can accomplish the same intellectual achievements humans can. It could reason, strategize, plan, use judgment and common sense, and respond to and detect hazards or dangers. This type of artificial intelligence is much more capable than the AI that powers the cameras in our smartphones, drives autonomous vehicles, or completes the complex tasks we see performed by ChatGPT. ... AGI could change our world, advance our society, and solve many of the complex problems humanity faces, to which a solution is far beyond humans' reach. It could even identify problems humans don't even know exist. "If implemented with a view to our greatest challenges, [AGI] can bring pivotal advances in healthcare, improvements to how we address climate change, and developments in education," says Chris Lloyd-Jones, head of open innovation at Avande. ... AGI carries considerable risks, and experts have warned that advancements in AI could cause significant disruptions to humankind. But expert opinions vary on quantifying the risks AGI could pose to society.

How to avoid the 4 main pitfalls of cloud identity management

DevOps and Security teams are often at odds with each other. DevOps wants to ship applications and software as fast and efficiently as possible, while Security’s goal is to slow the process down and make sure bad actors don’t get in. At the end of the day, both sides are right – fast development is useless if it creates misconfigurations or vulnerabilities and security is ineffective if it’s shoved toward the end of the process. Historically, deploying and managing IT infrastructure was a manual process. This setup could take hours or days to configure, and required coordination across multiple teams. (And time is money!) Infrastructure as code (IaC) changes all of that and enables developers to simply write code to deploy the necessary infrastructure. This is music to DevOps ears, but creates additional challenges for security teams. IaC puts infrastructure in the hands of developers, which is great for speed but introduces some potential risks. To remedy this, organizations need to be able to find and fix misconfigurations in IaC to automate testing and policy management.

Why a DevOps approach is crucial to securing containers and Kubernetes

DevOps, which is heavily focused on automation, has significantly accelerated development and delivery processes, making the production cycle lightning fast, leaving traditional security methods lagging behind, Carpenter says. “From a security perspective, the only way we get ahead of that is if we become part of that process,” he says. “Instead of checking everything at the point it’s deployed or after deployment, applying our policies, looking for problems, we embed that into the delivery pipeline and start checking security policy in an automated fashion at the time somebody writes source code, or the time they build a container image or ship that container image, in the same way developers today are very used to, in their pipelines.” It’s “shift left security,” or taking security policies and automating them in the pipeline to unearth problems before they get to production. It has the advantage of speeding up security testing and enables security teams to keep up with the efficient DevOps teams. “The more things we can fix early, the less we have to worry about in production and the more we can find new, emerging issues, more important issues, and we can deal with higher order problems inside the security team,” he says.

Understanding Europe's Cyber Resilience Act and What It Means for You

The act is broader than a typical IoT security standard because it also applies to software that is not embedded. That is to say, it applies to the software you might use on your desktop to interact with your IoT device, rather than just applying to the software on the device itself. Since non-embedded software is where many vulnerabilities take place, this is an important change. A second important change is the requirement for five years of security updates and vulnerability reporting. Few consumers who buy an IoT device expect regular software updates and security patches for that type of time range, but both will be a requirement under the CRA. The third important point of the standard is the requirement for some sort of reporting and alerting system for vulnerabilities so that consumers can report vulnerabilities, see the status of security and software updates for devices, and be warned of any risks. The CRA also requires that manufacturers notify the European Union Agency for Cybersecurity (ENISA) of a vulnerability within 24 hours of discovery. 

Conveying The AI Revolution To The Board: The Role Of The CIO In The Era Of Generative AI

Narratives can be powerful, especially when they’re rooted in reality. By curating a list of businesses that have thrived with or invested in AI—especially those within your sector—and bringing forth their successful integration case studies, you can demonstrate not just possibilities but proven success. It conveys a simple message: If they can, so can we. ... Change, especially one as foundational as AI, can be daunting. Set up a task force to outline the stages of AI implementation, starting with pilot projects. A clear, step-by-step road map demystifies the journey from our current state to an AI-integrated future. It offers a sense of direction by detailing resource allocations, potential milestones and timelines—transforming the AI proposition from a vague idea into a concrete plan. ... In our zeal to champion AI, we mustn’t overlook the ethical considerations it brings. Draft an AI ethics charter, highlighting principles and practices to ensure responsible AI adoption. Addressing issues like data privacy, bias mitigation and the need for transparent algorithms proactively showcases a balanced, responsible approach.

Chip industry strains to meet AI-fueled demands — will smaller LLMs help?

Avivah Litan, a distinguished vice president analyst at research firm Gartner, said sooner or later the scaling of GPU chips will fail to keep up with growth in AI model sizes. “So, continuing to make models bigger and bigger is not a viable option,” she said. iDEAL Semiconductor's Burns agreed, saying, "There will be a need to develop more efficient LLMs and AI solutions, but additional GPU production is an unavoidable part of this equation." "We must also focus on energy needs," he said. "There is a need to keep up in terms of both hardware and data center energy demand. Training an LLM can represent a significant carbon footprint. So we need to see improvements in GPU production, but also in the memory and power semiconductors that must be used to design the AI server that utilizes the GPU." Earlier this month, the world’s largest chipmaker, TSMC, admitted it's facing manufacturing constraints and limited availability of GPUs for AI and HPC applications. 

NoSQL Data Modeling Mistakes that Ruin Performance

Getting your data modeling wrong is one of the easiest ways to ruin your performance. And it’s especially easy to screw this up when you’re working with NoSQL, which (ironically) tends to be used for the most performance-sensitive workloads. NoSQL data modeling might initially appear quite simple: just model your data to suit your application’s access patterns. But in practice, that’s much easier said than done. Fixing data modeling is no fun, but it’s often a necessary evil. If your data modeling is fundamentally inefficient, your performance will suffer once you scale to some tipping point that varies based on your specific workload and deployment. Even if you adopt the fastest database on the most powerful infrastructure, you won’t be able to tap its full potential unless you get your data modeling right. ... How do you address large partitions via data modeling? Basically, it’s time to rethink your primary key. The primary key determines how your data will be distributed across the cluster, which improves performance as well as resource utilization.

AI and customer care: balancing automation and agent performance

AI alone brings real challenges to delivering outstanding customer service and satisfaction. For starters, this technology must be perfect, or it can lead to misunderstandings and errors that frustrate customers. It also lacks the humanised context of empathy and understanding of every customer’s individual and unique needs. A concern we see repeatedly is whether AI will eventually replace human engagement in customer service. Despite the recent advancements in AI technology, I think we can agree it remains increasingly unlikely. Complex issues that arise daily with customers still require human assistance. While AI’s strength lies in dealing with low-touch tasks and making agents more effective and productive, at this point, more nuanced issues still demand the human touch. However, the expectation from AI shouldn’t be to replace humans. Instead, the focus should be on how AI can streamline access to live-agent support and enhance the end-to-end customer care process. 

How to Handle the 3 Most Time-Consuming Data Management Activities

In the context of data replication or migration, data integrity can be compromised, resulting in inconsistencies or discrepancies between the source and target systems. This issue is identified as the second most common challenge faced by data producers, identified by 40% of organizations, according to The State of DataOps report. Replication processes generate redundant copies of data, while migration efforts may inadvertently leave extraneous data in the source system. Consequently, this situation can lead to uncertainty regarding which data version to rely upon and can result in wasteful consumption of storage resources. ... Another factor affecting data availability is the use of multiple cloud service providers and software vendors. Each offers proprietary tools and services for data storage and processing. Organizations that heavily invest in one platform may find it challenging to switch to an alternative due to compatibility issues. Transitioning away from an ecosystem can incur substantial costs and effort for data migration, application reconfiguration, and staff retraining.

The Secret of Protecting Society Against AI: More AI?

One of the areas of greatest concern with generative AI tools is the ease with which deepfakes -- images or recordings that have been convincingly altered and manipulated to misrepresent someone -- can be generated. Whether it is highly personalized emails or texts, audio generated to match the style, pitch, cadence, and appearance of actual employees, or even video crafted to appear indistinguishable from the real thing, phishing is taking on a new face. To combat this, tools, technologies, and processes must evolve to create verifications and validations to ensure that the parties on both ends of a conversation are trusted and validated. One of the methods of creating content with AI is using generative adversarial networks (GAN). With this methodology, two processes -- one called the generator and the other called the discriminator -- work together to generate output that is almost indistinguishable from the real thing. During training and generation, the tools go back and forth between the generator creating output and the discriminator trying to guess whether it is real or synthetic. 

Quote for the day:

''You are the only one who can use your ability. It is an awesome responsibility.'' -- Zig Ziglar

No comments:

Post a Comment