Artificial intelligence is rapidly becoming a core part of professional decision-making and our personal lives.

Presently, one of the greatest concerns facing AI solutions is termed overcompensation. “Overcompensating AI” means that when inappropriate prompts are sent to the AI engine, the system may produce overexaggerated responses or the funneling of the wrong dimensions, which generates an inappropriate solution. Issues like overcompensation highlight “the challenges of building and deploying AI systems that are both effective and responsible in direction,” Google says.

This article is a follow-up to those discussed in my April column, “Uncorking AI.” The purpose of this extension is to alert potential users that “false” outputs can be created from excessively biased or wrong data models that could have been generated before the current user started their prompts or sessions. AI learns by succession: It is trained by previous data sets composed of varying inputs and solutions it generates.

In addition to the examples herein, we’ll also look at practical solutions commercially available now and used today by many businesses, including the media.

Note that users of any AI should always be aware that uncertainty and challenges should always be expected, and thoughtful user criticism is essential to the output generated by any GenAI product regardless of the application.

Every provider of AI will strive to be “perfect” in its answers to prompts, inputs and inquiries of its systems. But this drive toward perfection might result in what is referred to as “overfitting,” that is, the solution reaching a point where the model it creates becomes excessively specialized.

Drive for Perfection and Automation Bias
This action may result in what the industry sometimes calls “hallucination.” In such cases, the details of products in their model may be camouflaged (i.e., hidden) to the user, and later the data model may be found to contain training data that, in turn, causes output solutions to perform poorly on new or unforeseen inputs (prompts).

When decisions rely too much on AI, there becomes a safety factor (i.e., an “overreliance”) that can stem from factors like “automation bias.” When users favor AI-generated information or lack a thorough understanding of the system limitations, studies have shown that people may find the AI suggestions sufficient or satisfactory and may tend to go outside their own judgment—even in high-stakes situations.

Human Evaluation vs. AI Observations
AI is great at objective assessments. It can easily filter and sort information based on metrics, which helps streamline specific tasks. However, when subjective judgment is required, human intuition is still superior.

In trying to reach “perfection,” the models and the outputs produced from those models can lead to “overfitting.” In other words, when or if a model becomes overly specialized and dependent solely on its training data—without a broad enough set of data points—it struggles to react appropriately to new or unexpected inputs from the user.

When the system overcompensates or has an overreliance bias to its data sets or models, implementations further downstream can result in continual misguidance to the outputs it provides. Thus, any AI system requires a balanced approach combining capabilities with human oversight and critical thinking.

Augmentation Tool
Research supports the idea that AI should best act as an augmentation tool. Reasoning by humans should routinely be incorporated into decision-making instead of simply replacing it. This process further “trains” the AI model while protecting decisions appropriately. Through Objective Assessments, which utilize AI, organizations can still free up human capacity for less-intensive decisions built typically on excessive efforts.

Subjective judgment may sometimes be required, in which case, human intuition is still superior—at least until such time that the computer solutions can better mimic the human brain in function and in performance, something that is still likely a decade or more away.

When properly navigated, AI helps people to make informed decisions without depending on subject-matter expertise. A collaborative approach involves experts who can critically evaluate AI “suggestions” instead of just accepting them without further analysis as “true without question.”

AI tools are constantly being created by leading software providers; without elaborating too far—some include Canva, Microsoft Co-Pilot, Grammarly and OpenAI’s ChatGPT (developed in November 2022). Such tools range from Large Language Models (LLM) to natural language, and to software generation and translation tools across multiple programming languages.

Software categories are quite broad—a small sampling of examples is in Fig 1. When choosing AI tool sets, it’s critical to consider factors like the specific use case, the software’s accuracy and performance, its data privacy and security measures, and the available training and support.

According to leading engineering and consulting firms, as much as 80% of an IT department’s budget can be consumed addressing outdated systems. Modernization is no longer just about efficiency, according to a report by Cognizant. Large-scale AI adoption will amplify the ability to rapidly respond to change.

Reaction time to market change can be improved and will allow the IT department to “innovate” instead of continuing to just “run” to stay up to date. To succeed in the AI age, we must, in turn, leverage technology and “overcome the tech debt that hampers innovation.”

Playbook for Growth and Innovation
The world’s growth and marketing playbooks are currently being rewritten. Marketing is now using AI “to do” and not to just “think.” To reach such goals with AI, an understanding of AI Optimization (AIO) will become essential to ranking on search engines.

“Ranking” refers to the position of a webpage in the search engine results pages (SERPs) when a user enters a query. Ranking allows for increased visibility, more traffic to the web pages, and the potential for conversions—that is, more users taking desired actions such as purchases or engaging of a new/improved service. AI is extremely useful in helping a website developer improve results by supporting the best practices for search engine optimization.

Defining, in part, what a “growth playbook” means is that there is now a fundamental shift in how today’s dynamic business environment is going to achieve and sustain growth. Businesses must now rethink and adapt their methodologies (Fig. 2). This goes well beyond what a few years ago was termed a “digital transformation”—as it means tools created for growth—especially AI—are now routine and these tools are now necessary to stay ahead of the curve.

Agentic AI
Businesses are now shifting to an agent-centric operations model that will, in turn, collapse their operational and data silos, while moving toward designing systems that think, act and scale with intelligence. AI has become essential to this rapidly changing business model.

Streamlining the operational model requires shifting from individual (siloed) functions and/or departments to a set of consolidated, semiautomated and anti-autonomous operations that provide continuous operational improvements, reduced time to obtaining the solution, and cutting costs while simultaneously making customer satisfaction and acceptance a “norm” rather than an “occasional” priority.

Companies such as Amazon, FedEx and even Starbucks have all found customer satisfaction improvements by employing “AI agents” to perform tasks which previously required individual humans in siloed departments to achieve the same tasks.

In the case of Starbucks, the company noted its “Deep Brew” program and the newly introduced “Green Dot Assist” are designed to optimize everything from inventory management and customer service to employee training and new product development. The company’s AI platform continually analyzes customer data, including purchase history, location, time of day and even weather patterns, to personalize recommendations, offers and rewards through the Starbucks app.

Media Literacy Couples AI Research
On the media side, mainstream organizations emphasize the importance of media literacy and verifying sources in this new era of informational warfare—especially where weaponized AI can now produce convincing fake content at low cost. Furthermore, Stanford University has developed research tools that utilize AI to analyze cable news coverage patterns, including detecting faces, identifying figures, estimating demographics and analyzing topic trends. These analysis methods help identify biases and trends in news reporting, including patterns of interruptions in on-air discussions.

Currently, AI is used to analyze thousands of dialogues on cable news programs to better understand the nature of interruptions in political discussions. “Interruptions” are when a speaker is saying something and is cut off by someone else who goes on to express their own thing, leaving the former embittered. The psychology of such interruptions are then cataloged, analyzed and used to train others (e.g., reporters or producers); using AI to better inform, criticize and prepare an interviewer on how to mitigate the negative impacts of such interruptions.

These are just some of the myriad value propositions of AI in the business world. Stay tuned to this column in future issues on how to better understand and use AI in your business or media application.

Scalability is generally attached to the concept whereby a system can expand from one particular “size” to another. Often the perception is that the top dimension is undefined—bringing to mind: “Just how large can this system expand to?” If you think about organizations (e.g., eBay, Amazon), there probably is no perception of the end point. Yet when applied to media asset management (MAM), the limits may be perceived by the number of records the system can handle, a capacity bounded by storage (not including the cloud), or by the effectiveness of the database to manage, search and retrieve the assets when needed and in a reasonable time period.

All this is quite ambiguous, to say the least. So, let’s put another descriptive term in front of scalability: “seamless.” Sometimes this becomes part of a marketing effect and sometimes the term is built in reality. In the case of the cloud the actuality probably can’t be identified regardless of the application—because “everything” is seamless in the cloud.

“Seamless scaling,” widely used to infer a system’s ability to expand to some level, is not necessarily a new term. Applicable models for the architecture are typically applied to networks, IP video streaming, CDNs, data storage and to MAM solutions. And most of these need to scale “seamlessly.”

When used in the “architecture” scenario, resources—and their usage—are, hopefully, scaled linearly against the load placed on the system by its users. For example, in compute operations, the load could be measured in the amount of user traffic; it could be set against input/output operations (IOPS); or may be related to data volume and more.

PRICE/PERFORMANCE CURVE

Resources must be balanced with performance. Here scalability is about resource usage associated with a single unit of work. Scalability, in this model, is about how resource usage and costs change when units of work grow in quantity or size. Scalability then becomes the “shape of the price-performance ratio curve,” as opposed to its value at any one point in the curve.

A price-to-performance ratio typically refers to a system’s (or product’s) ability to deliver performance for the price, capability or desire to pay. This falls into the proverbial “open checkbook” model, where cost becomes no object. However, there is a point where throwing in a boatload of cash ceases to deliver the performance desired in the time required. These factors impact the shape of that price-performance curve. See Fig. 1, where “time” can also be “performance.”

LINEARITY AND LATENCY

“Linear scaling” is sometimes depicted as a “straight line” model, its slope being determined by multiple factors—speed, throughput, latency, etc. In reality, systems utilizing technology simply won’t sustain a linear scaling model ad infinitum. Linear models will usually run at a given slope up to the point they fall over, dramatically change slope, become unstable or turn into a curve (Fig. 2).

Whether in a cloud or an on-prem datacenter, one variable impacting scalability is latency. Low-latency performance for workloads must be balanced with consistency. When latency dramatically changes, the net-net performance value is lowered (Fig. 3). Latency change is countered based upon design, system size, architecture, and the demand placed upon other services sharing the same resources.

One metric in cloud computing is its ability to deliver scalable access to a large pool of computational, storage and network resources, commonly known as infrastructure-as-a-service (IaaS). With functional workflows for storage, asset management, and databases moving more into the cloud, services leveraging machine learning and artificial intelligence make better owner/operator sense when they’re not built on premises.

Before the factors of price-to-performance are put into the model, a reality check about value proposition should occur and before the checkbook is opened or the bank account runs dry. It’s easy to let an unconstrained operational model go “cloud-wild” without implementing a set of checkpoints that help decide whether or not the practicality of going down this path is returning the value needed to reach the intended goal.

While the system (MAM, storage or network) may be able to scale seamlessly and linearly to a point without notice or consequence, users need to clearly put binders around the system that meets the goals in the time needed and without bankrupting the farm. Think again; perhaps the workflow storage solution is better kept at home.

“Super scaling” has been attached to hyper-converged compute platforms for at least a decade. Super scaling evolved from converged architectures as databases and analytics—regardless of their user applications—continued to demand real-time analysis for the delivery of its information and best performance.

Practical growth in these spaces previously relied heavily on massively parallel processors set into a clustered structure. The inhibiting force to this approach, in other than cloud, was the ability to access the storage at an I/O rate that matches the power of the combined processors, without latency or choking.

In terms of basic storage statistics and specifications, the rotational speed of the HDD, areal density of the bits, and the ability for data to be placed (written) or removed (read) from the storage material itself is not the whole story. Findings show other factors constraining scalability and performance for the storage system.

Solid-state drives (SSD) built on flash technologies, have changed the storage-dimension from what it was when only HDDs were available. Today, modern applications continue to push the envelope of storage I/O, capacities and processing. Seamless scaling now occurs in multiple dimensions aimed at supporting demand, change, data growth and adaptive user/workflows, all the time driving the question “cloud or datacenter/on-prem.” For the enterprise, direct attached SSDs (DAS) relay data to and from its servers, which strive to support an ever-expanding requirement set of transaction processing, data analytics and more.

DEMISE OF LEGACY DAS

DAS is successful because of the way it connects flash SSDs via PCIe, and continues to be a mainstream choice spanning over a dozen years. Unfortunately, the DAS+SSD scalability limit is nearly on the doorstep, to be relieved only by the comparatively recent technologies of non-volatile memory express (NVMe). Nonetheless, new methodologies still hold the constraints DAS had from infancy. Technically, if you only added a “shared-storage” model, then the value-proposition for NVMe became significantly diminished. Advancements in bus and I/O speeds changes that perspective, incorporating the principles of clustering and parallel processing for the storage environment.

Highly parallel, scale-out clustered applications require low-latency, high-performance shared storage capabilities. The latest change developed to address this weakness is that of the now standardized NVMe over Fabric (NVMe-oF). Fabrics, such in Fibre Channel or SANs, are software-defined resource topologies shared through interconnecting switches.

While all-flash arrays (AFA) changed storage models forever, its use of PCIe SSDs showed performance limitations when directly employed in application servers. Resources in this model became under-utilized to say the least—according to some, the net storage utilization averages between 30% and 40% and lower, in some cases.

Another issue in performance scaling is consistency. Some services, when run on a clustered server environment, can bring applications to a crawl. As services perform snapshots, cloning or other actions commanding CPU cycles, their functions take resources away from storage management activities—slowing data I/O and transfers.

Coupled with excessive data movement, complex operations and poor performance from large-footprint silicon devices—the negative impact on storage management cycles means that storage I/O becomes uncontrollable or variable to the point where errors develop, read/write cycles fluctuate or data becomes unavailable or worse, corrupted.

Recent improvements in scalable architectures, including storage, are supplementing advancements in physical storage footprints improving the capabilities to deliver data at the rates needed for workflows such as ultra-high definition (UHD), HDR/SDR and even the potential for uncompressed, high bitrate IP-flows (ST 2110) on servers and virtual machines.

In future storage articles, the depths of NVMe-oF will be expanded. In the meantime, if you’re looking at a storage refresh, investigate the (relatively discreet) manufacturers who are leveraging these new bus and data management technologies.

Karl Paulsen is CTO at Diversified and a SMPTE Fellow. He is a frequent contributor to TV Technology, focusing on storage and workflows for the industry. Contact Karl atkpaulsen@diversifiedus.com.