Skip to main content

Category Archives: Generative AI

Industry 4.0+: The Bridge to Industry 5.0

Posted on by

When Industry 4.0 was first coined (circa 2011), AI was primarily “Deterministic” and “Predictive.” It was about classifying defects or predicting when a something might fail. Since then, Generative AI (GenAI) and Agentic AI has emerged and has done significant advances in terms of core technology components and its applications. In my discussions with some  industry colleagues on understanding where Industry 4.0 stands today with these AI advancements, the sense I got was there is a belief, albeit wrong, that Industry 5.0 has replaced Industry 4.0 due to these advancements. So, before we talk about what these AI advances do to Industry 4.0, first lets us first get the misunderstanding out of the way in one sentence – Industry 5.0 complements the existing Industry 4.0 paradigm, not replace it, by highlighting research and innovation as drivers for a transition to a sustainable, human-centric and resilient industrial model.

A helpful way to view this is as two distinct but integrated layers:

  • Industry 4.0 is the “Technology Layer”: It focuses on machine-to-machine communication, IIoT connectivity, and autonomous digitalization to drive efficiency.
  • Industry 5.0 is the “Value Layer”: It serves as the “mission statement” for manufacturing, providing an ethical and societal wrapper that gives the technology its purpose.

Industry 5.0 was formally codified by the European Commission (EC) in 2021 with learning from disruptions of Covid-19 pandemic (Ref 1).

This Value Layer transitions the industry from a “Human-Out-of-the-Loop” model to a “Human-at-the-Center” collaboration by prioritizing three core pillars:

  • Human-Centricity: Technology is designed to adapt to the worker, treating them as an “investment / asset” rather than a mere cost or resource.
  • Sustainability: The goal shifts from simple efficiency to “Net Positive” impact, incorporating circular economy principles and low-carbon manufacturing.
  • Resilience: Influenced by the 2020–2022 global supply chain shocks, this pillar prioritizes a factory’s ability to pivot quickly (agility) over simply chasing the lowest possible unit cost.

If anybody is interested in the Intellectual Ancestor of Industry 5.0 – it is Japan’s “Society 5.0” (Ref 3).

One thing to guard against is – many technology vendors popularized an earlier, narrower version of Industry 5.0.

  • The Version: In this view, Industry 5.0 is simply Human-Robot Collaboration. It’s the era of the “Cobot” (Collaborative Robot), where the precision of a machine meets the creativity/craftsmanship of a human.
  • Cautionary Note: If you hear a hardware vendor talking about 5.0, they are usually referring to this “Man + Machine” technical setup. If you hear a policymaker or C-suite executive talking about it, they are usually referring to the broader “Human-Centric/Sustainable” EC framework.

Comparison at a Glance

Feature    Industry 4.0    Industry 5.0 (EC Formal)
Core Driver    Technology & Efficiency    Value & Purpose
Primary Goal    Automation & Connectivity    Human Wellbeing & Resilience
Role of AI    To replace/optimize human tasks    To augment/empower human talent
Success Metric    OEE (Overall Equipment Effectiveness)    ESG (Environmental, Social, Governance)

Now let us get back to what does the advances in Generative AI (GenAI) and Agentic AI mean for Industry 4.0 as it is still the ‘Technology Layer’ today. From what you see in the industry, below are the key alignments and shifts that have occurred to accommodate this evolution. Each of the points below are larger subjects. The view below is high level. Most of the shifts and alignments below are either in very early prototypes or active research.

  1. From “Predictive” to “Generative” Digital Twins

In the original Industry 4.0 framework, a Digital Twin was a virtual replica that mirrored physical reality to monitor and predict (predictive mirrors).

  • The Shift: With GenAI, Digital Twins have become Simulation Engines. Instead of just saying “this machine will fail,” GenAI can generate 10,000 synthetic failure scenarios to train models where real-world data is scarce (addressing the “small data” problem in niche manufacturing).
  • Agentic Alignment: Efforts are moving toward Agentic Digital Twins. These are not just mirrors; they are “agents” that can autonomously run “what-if” simulations in the background and then execute a physical change in the factory via the PLC (Programmable Logic Controller) without human intervention – currently primarily seen in research and pilot deployments.
  1. The Evolution of “Interoperability” (Natural Language Integration)

Original i4.0 focused heavily on vertical and horizontal integration via rigid protocols (OPC UA, MQTT, etc.).

  • The Shift: GenAI has introduced Semantic Interoperability. The “language barrier” between a legacy ERP system and a modern MES (Manufacturing Execution System) is being bridged by LLMs that can translate unstructured data and code.
  • Impact: We no longer require a specialized data scientist to write SQL queries for a report; a floor manager can ask an “Agentic Orchestrator” in natural language: “Rebalance the assembly line for 15% higher throughput using the current available staff,” and the agent negotiates the parameters across systems.

However, production deployments must pair language interfaces with strong schema grounding, access controls, and transactional guarantees before trusting agentic orchestration to change plant state. So, this necessitates transactional safeguards, conflict resolution, and human oversight in early stage.

  1. Re-defining the “Cyber-Physical System” (CPS)

The original definition of a CPS was a mechanism controlled or monitored by computer-based algorithms.

  • The Shift: We are seeing the rise of Cognitive-Physical Systems (Cog-PS).
  • Non-Agentic GenAI: Used for “Copilots”—or example, assisting maintenance technicians by synthesizing thousands of pages of manuals into a 3-step repair guide.
  • Agentic AI: These systems now possess multi-step reasoning. An agentic robot doesn’t just stop when it sees an obstacle (Traditional AI); it reasons through the delay, recalculates its path, notifies the next station of the delay, and adjusts its own speed to catch up.

But widespread closed‑loop autonomy requires staged validation, governance, and deterministic fallback. Agentic CPS are real and promising, but the path to safe, scalable autonomy is staged: build trust with copilots, validate with simulations and pilots, then expand agentic authority under strict governance.

  1. Convergence of Industry 4.0 and Industry 5.0

The emergence of GenAI has accelerated the transition to Industry 5.0, which adds “Human-Centricity” and “Resilience” to the efficiency of 4.0. This bridge can be referred to as Industry 4.0+, a reflection of how industries are extending beyond foundational digitalization to achieve higher autonomy, agility, and intelligence. A high-level view of differences is below.

Dimension

    Industry 4.0 (Traditional AI)

   Industry 4.0+ (GenAI / Agentic)
Worker Role    Operator of automated systems.    Collaborator with AI “Co-pilots.”
Optimization    Localized efficiency (OEE).    System-wide “Reasoning” and “Self-Healing.”
Design    CAD-based, human-led.    Generative Design (AI creates 100 iterations).
Maintenance    “If X, then Y” (Predictive).    “X happened, I’ve ordered parts and rescheduled.”

Industry 5.0 emphasizes human‑machine collaboration, neuroergonomic design, and operator empowerment—outcomes that GenAI copilots (AR guidance, SOP summarization, personalized dashboards) directly enable. Combining generative simulation, RAG grounding (Ref 4), and agentic orchestration (Ref 5) lets factories reason across supply, scheduling, and maintenance to self‑heal after disruptions, aligning with the complementary nature of 4.0 and 5.0.

  1. Updated Standards and Frameworks

Organizations like Platform Industrie 4.0 and NIST have begun updating their reference architectures (like RAMI 4.0) to include these new “Agentic” layers:

  • ISO/IEC 42001 (2023/2024): A new management system standard for AI that specifically addresses the governance needed for autonomous agents in industrial settings.
  • Synthetic Data Protocols: New frameworks are emerging to validate “Synthetic Data” generated by AI to ensure it doesn’t lead to “Model Collapse” in industrial safety systems.
    Note: For concept of Model Collapse, refer to my blog ‘Synthetic Data: A Double-Edged Sword?’ (Ref 2)

Many use the term “Industry 4.0+” mentioned above to bridge the gap between what industry was promised in 2011 and what is possible today. Below are some reasons why this informal term exists:

  1. The “Implementation Gap”

The original Industry 4.0 framework was highly theoretical. Many companies found that the “jump” to a fully autonomous, dark factory was too large.

  • The Term’s Purpose: Practitioners began using “4.0+” to describe the pragmatic, incremental upgrades to the existing 4.0 roadmap. It usually signifies the integration of Generative AI and Edge Computing into a system that was originally designed largely  for IIoT and Predictive Maintenance.
  1. Technical Maturity vs. Philosophical Shift

The reason we don’t have a formal “4.0+” definition is that the industry chose to move the “version number” to 5.0 to signal a change in intent, not just technology.

  • Industry 4.0+ is seen as a Vertical Upgrade: Better algorithms, faster 5G/6G, more powerful Agentic AI. It’s about doing the same things (efficiency, throughput) much better.
  • Industry 5.0 is a Horizontal Expansion: It adds new dimensions like worker well-being, CO2 neutrality, and supply chain resilience.
  1. How Different Groups Use “4.0+”

Because it isn’t formal, the definition potentially shifts depending on who you are talking to:

Player

   Their version of “Industry 4.0+”
Software Vendors    Usually means “Our software now has a GenAI Copilot.”
Academics    Refers to “Cyber-Physical-Social Systems”—the math of 4.0 meeting human behaviour.
System Integrators    Refers to “Brownfield 4.0″—upgrading legacy 3.0 machines with 4.0 intelligence using modern AI wrappers.

The best way to ensure that the vendors are really talking about solutions that have accommodated the recent AI technology advancements and not get blindsided by ‘+’, is to look at the AI Agency Level:

  • Level 1: AI observes / predicts / prescribes (Traditional 4.0). Based on patterns programmed. Very deterministic.
  • Level 2: AI suggests (Copilots/Non-agentic GenAI). Cognitive, knowledge driven, multidimensional
  • Level 3: AI executes & negotiates (Agentic 4.0+). Exhibits reasoning and multi-step autonomy.

These levels convey that the movement from Industry 4.0 to 5.0 as moving from “Smart Manufacturing” to “Thoughtful Manufacturing” as many in industry have started calling it.

In a separate blog post, we will explore how Industry 4.0 product vendors are responding to 4.0+ opportunity and 5.0 requirements.

References:

  1. Industry 5.0 , Towards a sustainable, human-centric and resilient European industry: https://op.europa.eu/en/publication-detail/-/publication/468a892a-5097-11eb-b59f-01aa75ed71a1/
    and
    https://op.europa.eu/en/publication-detail/-/publication/aed3280d-70fe-11eb-9ac9-01aa75ed71a1/language-en
  2. Synthetic Data: A Double-Edged Sword?’: https://aithoughts.org/synthetic-data-a-double-edged-sword/
  3. Industry 5.0, seriously?: https://investigationsquality.com/2025/05/31/industry-5-0-seriously/#:~:text=While%20Industry%204.0%20focused%20primarily,termed%20an%20%E2%80%9CImagination%20Society%E2%80%9D.
  4. Grounding AI Responses in Factual Data: https://medium.com/@minh.hoque/retrieval-augmented-generation-grounding-ai-responses-in-factual-data-b7855c059322
  5. Agentic orchestration: https://www.uipath.com/ai/what-is-agentic-orchestration

GenAI Adoption – Challenges in Manufacturing Enterprise

Posted on by

While discussions have been ongoing regarding the use of fine-tuned Large Language Models (LLMs) for specific enterprise needs, the high cost associated with cloud-based LLMs, including subscription fees and API usage charges, is becoming increasingly evident. This cost barrier has been a major hurdle for many enterprises seeking to transition GenAI-powered solutions from pilot programs to production environments.

Conversely, Small Language Models (SLMs) appear to be a more suitable option for businesses seeking specialized applications due to their lower cost and resource requirements. Enterprises typically operate with constrained budgets. Piero Molino (Chief Scientific Officer & Cofounder of Predibase, creator of Ludwig, formerly of Uber’s AI Lab), predicts that SLMs will be a major driver of enterprise adoption in 2024 due to their attractive financial proposition.

But within the enterprises sector, Manufacturing Enterprises will likely to be one of the slowest adopters of GenAI in their operations. Especially Medium and Small manufacturers. Let us explore the reasons because the combination of Industry 4.0’s data collection and connectivity with GenAI’s analytical and generative capabilities has significant potential to transform manufacturing into a more autonomous, intelligent, and efficient system.

Business Hurdles

Cost

The high cost of Large Language Models (LLMs) is a major hurdle for their adoption in manufacturing. LLMs require massive computing power for training and inference, making the enterprises reliant on cloud providers. However, cloud provider fees can scale rapidly with model size and usage, and vendor lock-in can be a concern.

Small Language Models (SLMs) offer a potential solution. Their lower computational footprint makes on-premises deployment a possibility. However, implementing SLM requires expertise in machine learning and LLM training, which some enterprises may lack. Hiring additional staff or finding a vendor with this expertise is an option, but maintaining an SLM on-premises can be complex and requires significant IT infrastructure.

For many manufacturing enterprises, the complexity and cost of on-premises SLM maintenance might outweigh the benefits of reduced cloud costs. This could lead them back to cloud based SLMs, landing them where they started.

Security Concerns

Security concerns around data privacy are a major hurdle for manufacturing companies considering both external vendors and cloud adoption. Usually Medium and Small manufacturers have viewed cloud with apprehension.

Change Management

Implementing Generative AI (GenAI) solutions can necessitate significant modifications to existing manufacturing software and may require changes to current processes. While change management might be straightforward for greenfield projects (entirely new systems), most implementations will be brownfield projects (upgrades to existing systems). Manufacturers are understandably hesitant to disrupt well-functioning manufacturing processes unless there’s a compelling reason. Therefore, a robust business case and a well-defined plan for minimizing disruption during change management are crucial.

Technical Hurdles

Data Challenges

GenAI models require large amounts of clean, labelled data to train effectively. Manufacturing processes can be complex and generate data that is siloed, inconsistent, or proprietary. So, unless there are existing Observability solutions which has captured sensor telemetry over period, the manufacturer cannot directly introduce a GenAI solution. Additionally, companies may be hesitant to share this data with external vendors.

Integration Complexity

Integrating GenAI solutions with existing manufacturing systems can be complex and require expertise in both AI and manufacturing technologies. Vendors may need to have experience working with similar manufacturing systems to ensure a smooth integration. Existing vendors may have to be roped in for the integration, which would incur additional cost. Integration governance could become complex.

Lack of Standardization

The field of GenAI is still evolving, and there is a lack of standardization in tools and techniques. This can make it difficult for companies to evaluate and select the right vendor for their needs.

Accuracy

SLMs are likely less susceptible to hallucination and bias compared to LLMs. SLMs are trained on a smaller amount of data, typically focused on a specific domain or task. SLMs have a simpler architecture compared to LLMs. Hence, they are less prone to situations where the model invents information or connections that aren’t there.

Data Quality Still Matters. Even with a smaller dataset, bias can still be present if the training data itself is biased. Bias, in case manufacturing systems, is about plant-shift based bias, machine life bias, role importance bias, vendor bias etc. Bias can also start building up through feedback loop from new production output.

Less Established Tools and Expertise

There are fewer established tools and frameworks specifically designed for SLMs compared to LLMs. Finding experts with experience in implementing SLM-based GenAI solutions might be more challenging.

Conclusion

What you will notice is that though there is cost reduction by using SLM instead of LLM, the challenges and hurdles remain almost the same. The hesitation that existed in Manufacturers for LLM based solutions remains for SLM based solutions. In many cases preventing the manufacturers moving from pilot to production. That hesitation needs to be tackled on priority basis to unlock the potential of SLM for the future of smart manufacturing.

Gen AI adoption: Is your budget ready?

Posted on by

As the adoption of Generative AI in the enterprise accelerates, one question that will be on Management’s mind: “What does AI cost?” The answer, like most things in business, is nuanced: it depends on the specific needs of the enterprise.

For a rough estimates, you can look at comparable businesses. For example, small enterprises with limited budgets might begin with AI-powered chatbots to automate customer support, freeing up existing staff for more complex tasks. But leaving at rough estimates is not a good approach.

Underestimating the importance of proper budgeting for adopting and operationalizing AI in the enterprise can be disastrous. A cautionary tale comes from cloud adoption, where unforeseen costs have triggered an exodus of businesses from cloud back to on-premises infrastructure.

Many sources, like those in Reference 1, meticulously list and explain the various cost factors involved. In a diagram here, these costs have been mapped onto different stages of Generative AI adoption. I haven’t elaborated on all stages in the diagram because some warrant their own detailed illustration.

Let us look at why a particular cost matters in the indicated stage. The assumption here is that the enterprises aim to fully implement and manage Gen AI themselves. The understanding would need little tweaking, but would hold good even when the enterprise decides to partially or completely outsources this activity.

Consultant Cost:

  • This is the cost of a consultant or consulting firm who will provide guidance and support throughout the AI adoption process.
  • In the diagram where this cost is not indicated, it is minimal compared to other costs at that stage.

Talent Cost:

  • Primarily encompasses the costs associated with reskilling current staff and hiring new talent.
  • Exercise caution, as advised by Hugo Huang, regarding the specific skills and headcount required for both AI solution implementation and ongoing maintenance.
  • Meticulous planning and budgeting are essential to prevent cost overruns.
  • While not explicitly indicated in the diagram in some stages, staff costs are assumed to be integrated within other categories such as Software Development Cost, Data Preparation Cost, and Rollout costs.

Cloud Cost:

  • Initial cloud costs will arise during the training phase, gradually scaling up to the target level during Rollout.
  • Carefully anticipate and plan for these costs, which are distributed across multiple stages.
  • If opting for an on-premises setup instead of cloud-based infrastructure, accurately factor in the equivalent costs.
  • Transitioning to on-premises infrastructure may necessitate a comprehensive review of existing infrastructure, potentially requiring additional efforts and budget allocation.

Inference Cost:

  • Initial inference costs will begin during training, escalating significantly during the three stages of Rollout. During steady state operations, this will be a major contributor to the on-going cost.

Data Preparation Cost:

  • Encompasses costs associated with data scientists, data analysts, and computing infrastructure (either cloud-based or on-premises).
  • Covers tasks such as cleansing, organizing, processing, and labelling data before it’s ready for training.
  • Expect a considerable time and money investment for this stage.
  • Additional costs may arise for implementing scalable and efficient data storage and data management systems.

Software Development Cost:

  • Involves costs related to building and testing applications that facilitate user interaction with the deployed GenAI solution.
  • Includes expenses for IT talent, licenses, and necessary infrastructure.

Fine-Tuning Cost:

  • Accounts for costs of personnel and infrastructure.
  • If synthetic data generation is planned using the trained model, include the related costs as well – such as personnel cost, inference cost, cloud cost etc..
  • Budget for this cost only if Fine-Tuning is part of the strategy.

Prompt Engineering Cost:

  • Allocate a budget for this cost if Prompt Engineering is chosen instead of, or in conjunction with, Fine-Tuning.
  • Primarily consists of costs associated with trained personnel.

Integration Cost:

  • This is for the cost of integrating newly built solutions with existing systems to ensure seamless user experience.
  • Involves the time and expertise of staff who manage these existing systems, even if not directly involved in GenAI implementation.
  • May necessitate changes to existing systems, requiring additional budget allocation.

Operations Cost:

  • Covers costs associated with the deployment and ongoing maintenance of the entire solution.

HBR’s Hugo Huang suggests management strategies for CEO/CIO for controlling costs. The CEO/CIO will constitute teams to carry out the GEN Ai adoption. While reviewing and signing-off the budgets / costs, the understanding of where the different costs occur will help.

Maryam Ashoori, in her article gets to the nuts-bolts of how to calculate different costs. A combination of similar approaches will help the teams constituted by CEO/CIO to make sure that costs well budgeted and under control.

To maintain competitive advantage, AI adoption in the enterprise is an unavoidable event. Cost estimation and control is one of the key pillars upon which a successful adoption of AI rests.

References:

  1. What CEOs Need to Know About the Costs of Adopting GenAI by Hugo Huang
    https://hbr.org/2023/11/what-ceos-need-to-know-about-the-costs-of-adopting-genai?ab=HP-latest-image-2
  2. Decoding the True Cost of Generative AI for Your Enterprise by Maryam Ashoori
    https://www.linkedin.com/pulse/decoding-true-cost-generative-ai-your-enterprise-maryam-ashoori-phd/

The ‘Ops’ in the GenAI World

Posted on by

The world of AI and its operational cousins can feel like an alphabet soup: AIOps, MLOps, DataOps, and now, GenAIOps. The key lies in understanding their distinct roles and how they can collaborate to deliver full potential of your Gen AI adoption and data investments.

Definitions

AIOps, which stands for Artificial Intelligence for IT Operations, is a rapidly evolving field that aims to leverage AI and machine learning to automate and optimize various tasks within IT operations.

MLOps, is a set of practices and tools that bring DevOps principles to the world of machine learning. It aims to automate and streamline the development, deployment, and maintenance of machine learning models in production.

DataOps, is essentially a set of practices, processes, and technologies that aim to improve the management and delivery of data products and applications. It borrows heavily from the DevOps methodology, applies it to the world of data.

GenAIOps, is the emerging field that applies the principles of AIOps, DataOps and MLOps to the specific challenges of managing and optimizing Generative AI systems.

Key Activities and Benefits

The table below captures the key objectives, activities and benefits of these ‘Ops’ areas.

Area Key Objectives Main Activities Benefits
AIOps Optimize AI infrastructure and operations ·    Automate manual tasks (incident detection, root cause analysis, remediation)
·    Improve monitoring and analytics (AI-powered analysis of IT data)
·    Proactive prediction and prevention (issue prediction from historical data)
·    Enhance collaboration and decision-making (unified platform for IT teams) 		·    Reduced downtime and costs
·    Improved AI performance
·    Faster problem resolution
·    More informed decision-making
MLOps Ensure efficient and reliable ML lifecycle ·    Automate ML pipeline (data pre-processing, training, deployment, monitoring)
·    Foster collaboration and communication (break down silos between teams)
·    Implement governance and security (compliance, ethical guidelines) 		·    Faster time to market for ML models
·    Increased model accuracy and reliability
·    Improved model governance and compliance
·    Reduced risk of model failures
DataOps Improve data quality, availability, and accessibility ·    Automate data pipelines (ingestion, transformation, delivery)
·    Implement data governance and quality control (standardization, validation)
·    Monitor data quality and lineage 		·    Improved data quality and trust
·    Better decision-making
·    Increased data accessibility and efficiency
·    Reduced data-related errors
GenAIOps Streamline and automate generative AI development and operations ·    Automate Generative AI pipelines (data preparation, training, output generation)
·    Monitor and manage Generative AI models (bias detection, remediation)
·    Implement governance and safety controls (bias mitigation, explainability tools)
·    Optimize resource allocation and cost management
·    Facilitate collaboration and communication 		·    Faster development and deployment of generative AI applications
·    Improved innovation and creativity
·    Efficient management of generative AI models
·    Reduced risk of bias and ethical issues in generative AI outputs

Comparative view

Because implementing GenAIOps would mostly require deploying MLOPs, DataOps and AIOPs also, it would be worthwhile to analyze distinctions and overlaps.

AIOps and MLOps

One uses AI, while the other applies DevOps principles.

AIOps:

  • Focus: Applying AI to improve IT operations as a whole.
  • Goals: Automate tasks, improve monitoring and analytics, predict and prevent issues, enhance collaboration and decision-making.
  • Examples: Using AI to detect network anomalies, automate incident resolution, or predict server failures.

MLOps:

  • Focus: Operationalizing and managing machine learning models effectively.
  • Goals: Automate the ML pipeline, deploy and monitor models in production, optimize performance, and ensure reliable and scalable operation.
  • Examples: Automating data pre-processing for model training, continuously monitoring model accuracy and bias, or automatically rolling back models when performance degrades.

Key Differences:

  • Scope: AIOps is broader, focusing on all aspects of IT operations, while MLOps is specifically about managing ML models.
  • Approach: AIOps uses AI as a tool for existing IT processes, while MLOps aims to fundamentally change how ML models are developed, deployed, and managed.
  • Impact: AIOps can improve the efficiency and reliability of IT operations, while MLOps can accelerate the adoption and impact of ML models in real-world applications.

Overlap and Synergy:

  • There is some overlap between AIOps and MLOps, especially in areas like monitoring and automation.
  • They can work together synergistically: AIOps can provide data and insights to improve MLOps, and MLOps can develop AI-powered tools that benefit AIOps.

So, while their core goals differ, AIOps and MLOps are complementary approaches that can together drive AI adoption and optimize both IT operations and ML models.

MLOps and GenAIOps

In the sense of focusing on operationalizing models, MLOps and GenAIOps share a similar core objective. Both aim to streamline the processes involved in deploying, monitoring, and maintaining models in production effectively. However, there are some key differences that distinguish them:

Type of models:

  • MLOps: Primarily focuses on managing traditional machine learning models used for tasks like classification, regression, or forecasting.
  • GenAIOps: Specifically deals with operationalizing Generative AI models capable of generating creative outputs like text, images, code, or music.

Challenges and complexities:

  • MLOps: Faces challenges like data quality and bias, model performance monitoring, and resource optimization.
  • GenAIOps: Grapples with additional complexities due to the unique nature of Generative AI, including:
    • Data diversity and bias: Ensuring diversity and mitigating bias in training data, as Generative AI models are particularly sensitive to these issues.
    • Explainability and interpretability: Providing tools and techniques to understand how Generative AI models make decisions and interpret their outputs, both for developers and users.
    • Ethical and regulatory considerations: Addressing ethical concerns and complying with relevant regulations surrounding Generative AI applications.

Tools and techniques:

  • MLOps: Tools for automating data pipelines, deploying models, monitoring performance, and managing resources might be sufficient.
  • GenAIOps: May require specialized tools and techniques tailored to address the unique challenges of Generative AI, such as:
    • Bias detection and mitigation tools: To identify and address potential biases in training data and model outputs.
    • Explainability frameworks: To facilitate understanding of how Generative AI models make decisions.
    • Content filtering and moderation tools: To ensure safe and responsible generation of outputs.

While both MLOps and GenAIOps share the general goal of operationalizing models, the specific challenges and complexities faced by Generative AI necessitate the development of specialized tools and practices within GenAIOps.

Collaboration:

  • AIOps and GenAIOps: These fields can coexist and complement each other within an organization. AIOps focuses on broader IT operations, while GenAIOps specifically addresses the unique challenges of managing Generative AI models. They can share data and insights to improve overall AI-driven decision-making and optimization.
  • MLOps and GenAIOps: While both focus on model operationalization, GenAIOps can be considered a specialized subset of MLOps that addresses the unique needs of Generative AI models. In organizations heavily invested in Generative AI, GenAIOps practices might naturally subsume the broader MLOps practices, ensuring tailored governance and operational efficiency for these advanced models.

Integration considerations:

  • Scope and Focus: Clearly define the scope of each field within your organization to ensure alignment and avoid overlap.
  • Tooling and Infrastructure: Evaluate whether existing MLOps tools can adequately support GenAIOps requirements or if specialized tools are needed.
  • Skill Sets: Foster cross-team collaboration and knowledge sharing to bridge gaps between different AIOps, MLOps, and GenAIOps teams. This is one of the most important considerations to keep operations cost down.

Summary and Future Outlook

  • AIOps and GenAIOps can coexist and collaborate for broader IT optimization and responsible Generative AI management.
  • GenAIOps can subsume MLOps practices in organizations with a strong focus on Generative AI, ensuring tailored governance and efficiency.
  • This convergence could lead to more comprehensive platforms and tools that address the entire AI lifecycle, from development to deployment, monitoring, and maintenance.

References

  1. What is AIOps? : https://www.ibm.com/topics/aiops
  2. What is MLOps and Why It Matters: https://www.databricks.com/glossary/mlops
  3. GenAIOps: Evolving the MLOps Framework: https://towardsdatascience.com/genaiops-evolving-the-mlops-framework-b0012f936379
  4. AI Project Management: The Roadmap to Success with AI, DataOps, and GenAIOps: https://www.techopedia.com/ai-project-management-the-roadmap-to-success-with-mlops-dataops-and-genaiops

GenAi & LLM: Impact on Human Jobs

Posted on by

I met an IT Head of a leading Manufacturing company in a social gathering. During our discussion, when he convincingly told me that current AI progress is destructive from the point of jobs done by humans and it’s going to be doomsday, I realized that many would be carrying a similar opinion, which I felt needs to be corrected.

A good starting point to understand impact of AI on jobs done by humans today is the World Economic Forum’s white paper published in September 2023 (Reference 1). It gives us a fascinating glimpse into the future of work in the era of Generative AI (GenAi) and Large Language Models (LLM). This report sheds light on the intricate dance between Generative AI and the future of employment, revealing some nuanced trends that are set to reshape the job market. Few key messages from the paper are below.

At the heart of the discussion is the distinction between jobs that are ripe for augmentation and those that face the prospect of automation. According to the report, jobs that involve routine, repetitive tasks are at a higher risk of automation. Tasks that can be easily defined and predicted might find themselves in the capable hands of AI. Think data entry, basic analysis, and other rule-based responsibilities. LLMs, with their ability to understand and generate human-like text, excel in scenarios where the tasks are well-defined and can be streamlined.

However, it’s not a doomsday scenario for human workers. In fact, the report emphasizes the idea of job augmentation rather than outright replacement. This means that while certain aspects of a job may be automated, there’s a simultaneous enhancement of human capabilities through collaboration with LLMs. It’s a symbiotic relationship where humans leverage the strengths of AI to become more efficient and dynamic in their roles. For instance, content creation, customer service, and decision-making processes could see a significant boost with the integration of LLMs.

Interestingly, the jobs that seem to thrive in this evolving landscape are the ones requiring a distinctly human touch. Roles demanding creativity, critical thinking, emotional intelligence, and nuanced communication are poised to flourish. LLMs, despite their impressive abilities, still grapple with the complexity of human emotions and the subtleties of creative expression. This places humans in a unique position to contribute in ways that machines currently cannot. But here the unique ability of LLMs to understand context, generate human-like text, and even assist in complex problem-solving, positions them as valuable tools for humans.

Imagine a future where content creation becomes a collaborative effort between human creativity and AI efficiency, or where customer service benefits from the empathetic understanding of LLMs. Decision-making processes, too, could see a paradigm shift as humans harness the analytical prowess of AI to make more informed and strategic choices.

There is also creation of new type of jobs, emerging jobs as it is called. For example, Ethics and Governance Specialists is one such emerging job.

The said paper further nicely brings together a view of job exposure by functional area and by industry groups: ranked by exposure (augmentation and automation potential) across large number of jobs to give reader a feel of what is stated above.

In essence, the report paints a picture of a future where humans and AI are not adversaries but partners in progress. The workplace becomes a dynamic arena where humans bring creativity, intuition, and emotional intelligence to the table, while LLMs contribute efficiency, data processing power, and a unique form of problem-solving. The key takeaway is one of collaboration, where the fusion of human and machine capabilities leads to a more productive, innovative, and engaging work environment. So, as we navigate this evolving landscape, it’s not about job replacement; it’s about embracing the opportunities that arise when humans and LLMs work hand in virtual hand.

 

References:

1.      Jobs of Tomorrow: Large Language Models and Jobs, September 2023. A World Economic Forum (WEF) white paper jointly authored by WEF and Accenture. https://www3.weforum.org/docs/WEF_Jobs_of_Tomorrow_Generative_AI_2023.pdf

 

 

Domain and LLM

Posted on by

I am in total agreement with Morgan Zimmerman, Dassault Systems quote in TOI today.  Every industry has its own terminologies, concepts, names, words i.e Industry Language. He says even a simple looking word like “Certification” have different meanings in Aerospace vs life sciences.  He recommends use of Industry specific language and your own company specific language for getting significant benefit out of LLMs. This will also reduce hallucinations and misunderstanding.

This is in line with @AiThoughts.Org thoughts on Domain and company specific information on top of general data used by all LLMs.  Like they say in Real Estate, the 3 most important things in any real estate buy decision is “Location, Location and Location”.  We need 3 things to make LLMs work for the enterprise.  “Domain, Domain and Domain”.   Many of us may recall a very successful Bill Clinton Presidential campaign slogan. “The economy, Stupid”.   We can say “The domain, Stupid” as the slogan to make LLMs useful for the enterprises.

But the million-dollar question is how much it is going to cost for the learning updates using your Domain and company data?  EY published a cost of $1.4 Billion which very few can afford.  We need much less expensive solutions for large scale implementation of LLMs.

Solicit your thoughts. #LLM #aiml #Aiethics #Aiforindustry

L Ravichandran

Generative AI in Plant Maintenance

Posted on by

Today almost all manufacturing verticals are highly competitive making it necessary to avoid any breakdown of the equipment in manufacturing process. This has made the past practice of Reactive Maintenance unacceptable. Aiming to eliminate breakdowns has other benefits like improved employee motivation, reduction in opportunity costs, and reduction in production cost.

There are broadly 6 maintenance types, and they are indicated in order of maturity: Ref.1

1.       Reactive Maintenance: When it breaks, you fix it. This is where most of the manufacturers start. It results in emergency maintenance which is unintentional and consists of repairing and replacing equipment on a “fire-fighting” basis. Production loss is usual result.

2.       Preventive Maintenance: You schedule replacements ahead of time before parts break, usually at a regular interval.

3.       Usage-Based Maintenance: You replace material when the machine has been used a certain amount before they break. You change oil in the equipment after say usage of 5000 hours. It doesn’t matter if it takes you one month or one year to hit five thousand hours, the oil only needs to be replaced once it has been used to its potential and further use could cause degradation of other parts.

4.       Condition-Based Maintenance: You replace the parts when they seem like they are getting too worn out to continue to function appropriately. Measurement of condition of the parts can be manual where very frequent inspections are carried out or it can be continuous by using sensors attached to the equipment. This results in more usage for the money spent.

5.       Predictive Maintenance: Utilize historical data to make predictions about when a part will break and when to replace the parts based on these predictions, prior to them breaking. This usually utilizes IIoT (Industrial IoT) and utilizes, but not always, artificial intelligence and machine learning. But it still depends on managers to take actions, like creating work order, assigning technicians etc.

6.       Prescriptive Maintenance: Advanced data analysis methods are used to do more than predict failure points, but instead provides hypothetical outcomes in order to choose the best action that can be taken – to avoid or delay failures – prior to the failure, safety hazards, and quality issues arise. It automatically creates work orders. It requires no intervention from managers and oversees equipment on its own. Generative AI (Gen AI) is helpful here.

A manufacturer implements a combination of approaches as above, based on cost-benefit analysis. The two approaches benefit significantly by utilizing AI techniques are Predictive and Prescriptive Maintenance.

The three core systems that are connected with each other are Asset Management, Maintenance Management and Inventory Management.

1.       Asset Management System (AMS): maintains map of assets deployed and their characteristics. It monitors the wear and tear, hence remaining life of assets or its parts. It sends Work Orders as triggers for maintenance requirements to Maintenance Management System.

2.       Maintenance Management System (MMS): It acts on the Work Orders from AMS to generate activity plan, inventory allocation / ordering, technician scheduling, calendar management that is required to get the job done.

3. Inventory Management System (IMS): Stores the current inventory with its parts characteristics and vendor details. On trigger from MMS, it either allocates available parts from existing inventory or gets the part through its ordering process.

The infusion of Gen AI in Plant Maintenance consists of below three key aspects:

1.       Continuous Condition Monitoring

2.       Predicting failures

3.       Executing repairs / replacements

Continuous Condition Monitoring is primarily implemented by deployment of various sensors enabled by IIoT and availability of plantwide WiFi connectivity for feeding the real time inputs as Time Series data to AMS. For example, sensors are deployed on all motors to pick rotation, speed, temperature data and send it continuously to AMS. In some cases, this could be even vision (image) data from cameras. For example, in case of monitoring depth ‘roller grooves’ used to roll steel bars from the steel billets in Steel Industry. AMS consumes all these inputs.

Predicting Failures is typically done by AMS. The data from the Continuous Monitoring is usually fed as Time Series data to AMS. Using the Machine Learning part of AI domain, the Real Time data received is analyzed using models trained by historical data of sensors, usually in correlation with different product manufacturing campaigns, to anticipate potential breakdowns in equipment or their parts. The parts’ technical data from product vendor is also used.

For example, from vendor provided data on motor, the life expectancy of motor in number of hours at certain load is known to AMS from the IMS where all details of the motor are stored. The hours of running of a motor and the load at which it was running can be found out from the readings of current drawn. The AMS may pick up all the readings of current drawn from the Time Series data it receives from sensors / meters and calculates the hours of running and average load during that period.

Similarly, for each equipment, there may be different sensor readings which can be used to calculate the used life. Applying the learning of the trained model, insights are generated to detect developing defects before they become major problems, determine the remaining usable life (RUL) of the assets. AMS then generates Work Order as request for the maintenance along with the constraints like outer time limit before the maintenance must be done.

Executing repairs  / replacements is done by MMS. Based on the maintenance request, the MMS deduces material required, skill required, and work-shift calendar carries out below activities:

  • –          Receive Work Order to carry out the task and generate activity plan.
  • –          Assign a technician, based on required skills, individual availability through the calendar and plant’s holiday schedule. Put that as task in the technician’s calendar.
  • –          Book tools required for technician’s work.
  • –          Put the request for the required material into IMS to get it allocated or purchased and then allocated.
  • –          Update AMS, when job is done, with required details so that the monitoring can start again.

Opportunities

The amount of data available for training AI models is key to its ability to identify patterns and arriving at decisions. But obtaining large, labelled datasets can be challenging. Gen AI can be used to create new dataset matching the same underlying patterns as the original one. Such dataset can also be generated to bring in various conditions and failure scenarios that otherwise is not possible to capture with historical data alone. Availability of such large dataset ensures rigorous testing of prediction models, while mitigating bias in the model and enhancing quality of prediction.

With its ability to consume multi-modal inputs like sensor data, images and texts from manuals, camera inputs etc., it is possible to generate more comprehensive understanding of machine’s or part’s health fostering faster anomaly detection, better prediction, and accurate maintenance recommendation.

Gen AI can be used to create a Job Card and schedules for the repairs based not only on dimensions stated above, but also on analysing past performance of the technician for similar repairs, records of Mean Time Between Repairs etc.

So, Gen AI does not just predict a problem, but provides a solution. When a machine or its part shows signs of potential failure, Gen AI can look at a set of viable solutions and then generates a Work Order that ensures the most suitable fix.

No two similar machines or similar parts wear out similarly. Gen AI can generate different work orders and schedules based on real time data, corelating it with other influencing parameters. This ensures cost effectiveness in maintenance.

An interesting side effect of the ability of Gen AI to create a large set of synthetic data from a small set of actual data it its use as training tool. It can simulate a plethora of machinery failure scenarios to offer realistic training experience for technicians.

Challenges

The complexity involved in deploying Gen AI in Plant Maintenance required significant computing power. The natural choice becomes usage of Cloud based infrastructure. Hence safeguarding data privacy and security becomes paramount, as it involves sensitive equipment information and maintenance logs.

Conclusion

Gen AI brings in lot of improvement opportunities in Plant Maintenance through greater accuracy, efficiency, and reliability. The implementation exercise should take cognizance of challenges involved to make the adoption successful.

References:

1.       The different types of maintenance in manufacturing by Graham Immerman, MachineMetrics, 2020