Daily Tech Digest - January 10, 2025

Meta puts the ‘Dead Internet Theory’ into practice

In the old days, when Meta was called Facebook, the company wrapped every new initiative in the warm metaphorical blanket of “human connection”—connecting people to each other. Now, it appears Meta wants users to engage with anyone or anything—real or fake doesn’t matter, as long as they’re “engaging,” which is to say spending time on the platforms and money on the advertised products and services. In other words, Meta has so many users that the only way to continue its previous rapid growth is to build users out of AI. The good news is that Meta’s “Dead Internet” projects are not going well. ... Meta is testing a program called “Creator AI,” which enables influencers to create AI-generated bot versions of themselves. These bots would be designed to look, act, sound, and write like the influencers who made them, and would be trained on the wording of their posts. The influencer bots would engage in interactive direct messages and respond to comments on posts, fueling the unhealthy parasocial relationships millions already have with celebrities and influencers on Meta platforms. The other “benefit” is that the influencers could “outsource” fan engagement to a bot. ... “We expect these AIs to actually, over time, exist on our platforms, kind of in the same way that accounts do,” Connor Hayes, vice president of product for generative AI at Meta, said


Experts Highlight Flaws within Government’s Data Request Mandate Under DPDP Rules 2025

Tech Lawyer Varun Sen Bahl also points out the absence of an appellate mechanism for such ‘calls for information’ by the Central government, explaining that such an appeal process only extends against orders of the Data Protection Board. He explains, “This is problematic because it leaves Data Fiduciaries and Data Principals with no clear recourse against excessive data collection requests made under Section 36 read with Rule 22“. Bahl also notes that the provision lacks specific mention of guardrails like the European Union’s data minimisation principle under the General Data Protection Regulation (GDPR) while furnishing such information requests. ... Roy argues that the compliance burdens on Data Fiduciaries will increase and aggravate through sweeping requests and by invoking the non-disclosure clause. To explain, he cites the case of the Razorpay-AltNews situation in 2022, when the Government accessed the names and transaction details of the news platform’s donors via Razorpay ... To ensure that government officers and agencies don’t abuse this provision, Roy explains that “Fiduciaries must [as part of corporate governance] give periodic reports of the number of such demands”. Similarly, law enforcement and other agencies should also submit periodic reports of such requests to the Data Protection Board comprising details of cases where the non-disclosure clause is invoked.


How Edge Computing can Give OEMs a Competitive Advantage

Latency matters in warehouse automation too. Performing predictive maintenance on a shoe sorter, for example, could require real-time monitoring of actuators that do diversions every 40 milliseconds. Component-level computing power allows the system to respond to changing conditions with speed and efficiency levels that simply wouldn’t be possible with a cloud-based system. ... Edge components can also communicate with a system’s programmable logic controllers (PLCs), making their data immediately available to end users. Supporting software on the customer’s local network interprets this information, enabling predictive maintenance and other real-time insights while tracking historical trends over time. ... Edge technology enables you to build assets that deliver higher utilization to your customers. Much of this benefit comes from the greater efficiencies of predictive maintenance. Users have less downtime because unnecessary service is reduced or eliminated, and many problems can be resolved before they cause unplanned shutdowns. Smart components can also deliver more process consistency. Ordinarily, parts degrade over time, gradually losing speed and/or power. With edge capabilities, they can continuously adapt to changing conditions, including varying parcel weights and normal wear.


Have we reached the end of ‘too expensive’ for enterprise software?

LLMs are now changing the way companies approach problems that are difficult or impossible to solve algorithmically, although the term “language” in Large Language Models is misleading. ... GenAI enables a variety of features that were previously too complex, too expensive, or completely out of reach for most organizations because they required investments in customized ML solutions or complex algorithms. ... Companies need to recognize generative AI for what it is: a general-purpose technology that touches everything. It will become part of the standard software development stack, as well as an integral enabler of new or existing features. Ensuring the future viability of your software development requires not only acquiring AI tools for software development but also preparing infrastructure, design patterns and operations for the growing influence of AI. As this happens, the role of software architects, developers, and product designers will also evolve. They will need to develop new skills and strategies for designing AI features, handling non-deterministic outputs, and integrating seamlessly with various enterprise systems. Soft skills and collaboration between technical and non-technical roles will become more important than ever, as pure hard skills become cheaper and more automatable.


Is prompt engineering a 'fad' hindering AI progress?

Motivated by the belief that "a well-crafted prompt is essential for obtaining accurate and relevant outputs from LLMs," aggressive AI users -- such as ride-sharing service Uber -- have created whole disciplines around the topic. And yet, there is a reasoned argument to be made that prompts are the wrong interface for most users of gen AI, including experts. "It is my professional opinion that prompting is a poor user interface for generative AI systems, which should be phased out as quickly as possible," writes Meredith Ringel Morris, principal scientist for Human-AI Interaction for Google's DeepMind research unit, in the December issue of computer science journal Communications of the ACM. Prompts are not really "natural language interfaces," Morris points out. They are "pseudo" natural language, in that much of what makes them work is unnatural. ... In place of prompting, Morris suggests a variety of approaches. These include more constrained user interfaces with familiar buttons to give average users predictable results; "true" natural language interfaces; or a variety of other "high-bandwidth" approaches such as "gesture interfaces, affective interfaces (that is, mediated by emotional states), direct-manipulation interfaces


Building Resilience Into Cyber-Physical Systems Has Never Been This Mission-Critical

In our quest for cyber resilience, we sometimes—mistakenly—fixate on hypothetical doomsday scenarios. While this apocalyptic and fear-based thinking can be an instinctual response to the threats we face, it is not realistic or helpful. Instead, we must champion the progress, even incremental, that is achievable through focused, pragmatic measures—like cyber insurance. By reframing discussions around tangible outcomes such as financial stability and public safety, we can cultivate a clearer sense of priorities. Regulatory frameworks may eventually align incentives towards better cybersecurity practices, but in the interim, transferring risk via a measure like cyber insurance offers a potent mechanism to enhance visibility into risk mitigation strategies and implement better cyber hygiene accordingly. By quantifying potential losses and incentivizing proactive security measures, cyber insurance can catalyze a necessary, and overdue cultural shift towards resilience-oriented practices—and a safer world. We stand at a pivotal moment in American critical infrastructure cybersecurity. As hackers threaten to sabotage our vital systems for ransom, the financial damages ensued from incidents like Halliburton oblige us to stay alert and act proactively. 


Don't Fall Into the 'Microservices Are Cool' Trap and Know When to Stick to Monolith Instead

Over time, as monolith applications become less and less maintainable, some teams decide that the only way to solve the problem is to start refactoring by breaking their application into microservices. Other teams make this decision just because "microservices are cool." This process takes a lot of time and sometimes brings even more maintenance overhead. Before going into this, it's crucial to carefully consider all the pros and cons and ensure you've reached your current monolith architecture limits. And remember, it is easier to break than to build. ... As you see, the modular monolith is the way to get the best from both worlds. It is like running independent microservices inside a single monolith but avoiding collateral microservices overhead. One of the limitations you may have – is not being able to scale different modules independently. You will have as many monolith instances as required by the most loaded module, which may lead to excessive resource consumption. The other drawback is the limitations of using different technologies. ... When running a monolith application, you can usually maintain a simpler infrastructure. Options like virtual machines or PaaS solutions (such as AWS EC2) will suffice. Also, you can handle much of the scaling, configuration, upgrades, and monitoring manually or with simple tools. 


SEC rule confusion continues to put CISOs in a bind a year after a major revision

“There is so much fear out there right now because there is a lack of clarity,” Sullivan told CSO. “The government is regulating through enforcement actions, and we get incomplete information about each case, which leads to rampant speculation.” As things stand, CISOs and their colleagues must chart a tricky course in meeting reporting requirements in the event of a cyber security incident or breach, Shusko says. That means anticipating the need to deal with reporting requirements by making compliance preparation part of any incident response plan, Shusko says. If they must make a cyber incident disclosure, companies should attempt to be compliant and forthcoming while seeking to avoid releasing information that could inadvertently point towards unresolved security shortcomings that future attackers might be able to exploit. ... Given that clarity around disclosure isn’t always straightforward, there is no real substitute for preparedness, and that makes it essential to practise situations that would require disclosure through tabletops and other exercises, according to Simon Edwards, chief exec of security testing firm SE Labs. “Speaking as someone who is invested heavily in the security of my company, I’d say that the most obvious and valuable thing a CISO can do is roleplay through an incident.”


How adding capacity to a network could reduce IT costs

Have you heard the phrase “bandwidth economy of scale?” It’s a sophisticated way of saying that the cost per bit to move a lot of bits is less than it is to move a few. In the decades that information technology evolved from punched cards to PCs and mobile devices, we’ve taken advantage of this principle by concentrating traffic from the access edge inward to fast trunks. ... Higher capacity throughout the network means less congestion. It’s old-think, they say, to assume that if you have faster LAN connections to users and servers, you’ll admit more traffic and congest trunks. “Applications determine traffic,” one CIO pointed out. “The network doesn’t suck data into it at the interface. Applications push it.” Faster connections mean less congestion, which means fewer complaints, and more alternate paths to take without traffic delay and loss, which also reduces complaints. In fact, anything that creates packet loss, outages, even latency, creates complaints, and addressing complaints is a big source of opex. The complexity comes in because network speed impacts user/application quality of experience in multiple ways, ways beyond the obvious congestion impacts. When a data packet passes through a switch or router, it’s exposed to two things that can delay it.


Ephemeral environments in cloud-native development

An emerging trend in cloud computing is using ephemeral environments for development and testing. Ephemeral environments are temporary, isolated spaces created for specific projects. They allow developers to swiftly spin up an environment, conduct testing, and then dismantle it once the task is complete. ... At first, ephemeral environments sound ideal. The capacity for rapid provisioning aligns seamlessly with modern agile development philosophies. However, deploying these spaces is fraught with complexities that require thorough consideration before wholeheartedly embracing them. ... The initial setup and ongoing management of ephemeral environments can still incur considerable costs, especially in organizations that lack effective automation practices. If one must spend significant time and resources establishing these environments and maintaining their life cycle, the expected savings can quickly diminish. Automation isn’t merely a buzzword; it requires investment in tools, training, and sometimes a cultural shift within the organization. Many enterprises may still be tethered to operational costs that can potentially undermine the presumed benefits. This seems to be a systemic issue with cloud-native anything.



Quote for the day:

"The best leader brings out the best in those he has stewardship over." -- J. Richard Clarke

No comments:

Post a Comment