I came across this youtube video https://www.youtube.com/watch?v=PBdZi_JtV4c&t=1008s. Excerpts from Web summit discussion titled “Debunking the great AI Lie” by Noam Chomsky, Gary Marcus, and Jeremy Khan.

The topic itself is very catchy and as an evangelist for “AI for Enterprise”, I was intrigued and listened to the 32-minute clip. At the outset, the arguments about language complexities, human and language connections, and how pure data-driven tools like GPT3, etc. are not sufficient to emulate full language comprehension were compelling. Gary mentioned the lack of sufficient early inputs from language experts in the development of the AI model. Chomsky used a very strong statement (even though he said he is being nice to AGI) about GPT3 and other types of language model research. He emphatically said these research efforts neither advance engineering capabilities (like telescopes in astronomy) nor advance scientific research on languages. However, he also mentioned that due to his hearing problem, he was able to understand questions from Jeremy using the real-time voice-to-text transaction. Gary also felt that $billions of research on autonomous AI were also a waste of money and effort. Gary also admitted that while these data-based models get it right 75% of the time, they can never get it right the remaining 25%.

I do not agree with the statement that these research areas and the products are a total waste. Research on autonomous cars is being used every day when you drive any mid-size car around the world. Parking assist, reverse assist, and lane change alerts are standard features in most cars and are used by millions of people. Language translators are helping millions of tourists around the world to do basic communication in foreign destinations. Voice-to-text tools like Alexa, and Siri have become standard appliances in most households.

We will leave the philosophical questions raised by the gurus about language nuances, syntax-semantic connections, the importance of past memories, etc for later and get back to our theme at @AiThoughts.Org i.e. “AI for Enterprise”. I have already argued in my earlier blog about how Artificial Narrow Intelligence is sufficient for the next level of enterprise automation and performance enhancement. I can reiterate by saying “Narrow AI is Deep enough for Enterprises” and we welcome all research efforts towards AGI which will eventually help in making ANI models much more accurate and solve enterprise problems.

Just to recall from my earlier Blog, I talked about HR-Operations certified expert NAI model able to understand most of the employee’s queries on HR-Operations issues and invoke the right ERP applications to complete the transactions. Gary’s assessment of 75% AGI correctness may translate to maybe over 90% in a limited domain narrow intelligence world such as HR-Ops, Invoicing Ops, Billing Ops, etc. I feel the research should not worry about reducing the gap of 10% but create an egoless model which can quickly realize that the employee’s query is not understandable by itself and transfer the call to a human agent in 10% of cases. How quickly and how seamlessly the model does the switch is what makes for a superior customer experience.

Hip-Hip-Hooray for the AI researchers both in AGI and ANI. Each will feed into the other to make AI models more accurate and usable. Enterprises have a huge amount of open source and commercial technologies and models available in ANI to get the best performance and they should allocate sufficient budgets and accelerate AIML projects in 2023.

More later,

L Ravichandran

Many of you have seen a news item from MIT https://news.mit.edu/2022/machine-learning-university-math-0803. The headlines was “New Algorithm aces University level Maths questions”. Lots of talk on Algorithms passing the undergraduate course in Computer science but due to their non-person status degrees cannot be legally given by Universities etc. Technically, these Algorithms, even though look so intelligent are classified as ANI as they are experts in only one area i.e. Maths or Computer Science.

Our mission and thoughts in @AiToughts.Org is always on #AIforEnterprise and We feel that Narrow is good. @sivaguru already talked about power of mixing AIML algorithms with virtual assistants in achieving much higher levels of automation. Let me illustrate this with a HR example. Many organizations have attempted to build HR Ops chatbots to improve employee experience. However, all these efforts have not achieved the full desired goals and full scale roll out and cutting HR Ops cost is yet a reality.

Let us use the MIT example and build a Algorithm who can pass HR Ops field. I have looked at various HR Ops orientation courses in large enterprises and also HR Ops courses in various MBA courses. Let the algorithm learn about all the technical nuances of HR policies, implementation, work flows and various nuances such as mixing of long weekends with vacation leave, health insurance issues, induction and exit process and bottlenecks. On top of the algorithm which is HROPs-Certified, sits a chatbot with local speaking nuances in the deployed country and a customized personal assistant.

Let us do a simulated run of this new HPOPs Certified ANI system ; let us call it HRANI

Scenario 1

Time 630am Local time : Employee shift is 900am to 500PM Local time.

Employee Peter: HI HRANI , I am sick

HRANI : Looks at the time and deduces that employee is at home and will not be able to come to work. Personal assistant logs in to HR ERP system and finds out employee has sick leave available. Personal assistant also checks the calendar and sees many meetings where the employee is accepted.

Hi Peter, Sorry to hear that you are sick. I am sure you will get well soon. You have sick leave entitlement and I have applied for sick leave. I have also sent a e-mail to your Boss xxx and your reportees mentioning about your leave of absence today. BTW, you have many meetings scheduled and do you plan to take them from home or shall I cancel.

Peter : Hi HRANI, Please cancel all meetings except 1200PM one.

Hirani : Done, Peter. Take rest. Hope you get better soon.

Scenario 2

Time is 100PM local time.

Peter : HI HRANI, I am sick

HRANI : Hi Peter, sorry to hear about your sickness. We have a campus doctor to consult and also a sick bay for you to rest. Shall I arrange this for you?

Same 3 short words from Employee “I am sick” is handled by HRANI in two different ways like a HR Ops person will do!.

Scenario 3

Peter : HI HRANI, I want to take off with family next month for couple of weeks

HRANI : Great Idea Peter. You need a break. You have not taken a single day off for last 9 months. There is a 3 days long weekend coming up and you can take vacation around it from dd/mm/yy to dd/mm/yy. You are short of 1 day leave and I can write a special approval request to your manager and with your track record I am confident that this will be approved.

Peter: Please go ahead.

HRANI knows about leave nuances of combining with weekends also knows about work flow for exceptions and approval probabilities.

Scenario 4

HRANI : Hi Peter, Happy to hear about your promotion and salary raise. I can suggest few ways to do better tax planning. Would you be interested?

Peter : Yes.

HRANI : Your PF ( 401K) contribution is Rs. Xxxx and if you take it up to Rs. Yyyy and if you donate RS. XXX to some charities, then your tax bracket will come down a notch and you save bundle of money.

Peter: Thanks, HRANI. I will consider this suggestion.

HRANI knows abut salary components, local tax laws and exceptions and can do a what if analysis and proactively advise employees similar to a HR Ops person.

This is Narrow as this only knows about HR Ops area. I am Good with it. We can also have ANIs for Billing Ops, Account receivables Ops etc. which take up bulk of operational costs in enterprises.

Can these scenarios possible in an HRANI ? The current NLP technologies to understand natural written text is pretty good and it is not that difficult to build algorithms to understand the Intent and also the context from free format human text. The various HROps scenarios are all written down in new HR employee orientation books and can be easily taught to algorithms by supervised learning methods.

To conclude, ANIs can make a huge impact to enterprises. I am OK with Narrow!

One of the techniques popular among AI adoption and deployment initiatives is the use of Digital Twins.

While in its simplest form, they may be equated to simulations, there could be different levels of sophistication and applications.

In AI projects, particularly, Digital Twins can help accelerate the time to solution.

For example, in the real world, if one is modeling a prediction algorithm for failure recovery or preventive maintenance, gather actual performance data from physical devices may not only be time consuming, as one has to wait for a long time to reach some boundary conditions or gather sufficient data.

Using a digital twin, one may shoten that duration.

This approach would also be useful when, let us say, the device [an IoT end point?] is not easily accessible.

An extreme extension of this is when we can have digital assistants that can take some work load from us.

We would still want this assistant to be as close to our style and preferences in terms of how do things.

When I came across tools that generate articles, blog posts etc, wanted to try them out..

The article below has been generated by AI.

I used a tool called writesonic and chose to create this article in a guided manner.

I only had to type digital twin as the starting phase for the topic.

I answered a few questions to pick from 3-4 options suggested for the structure, style etc.. and bingo! This was generated!

Digital Twins: What They Are and Why You Need One

Image Source: FreeImages

‍

Digital twins are digital representations of physical objects. They are also known as digital avatars, virtual twins, or digital companion models. Digital twins are becoming increasingly important in various industries because of the rise of machine learning and artificial intelligence (AI). This article will explain what a digital twin is, why you need one, and its many different types.

What is a Digital Twin?

A digital twin is a virtual replica of a real-world asset that, as the name suggests, is held in digital form. Digital twins are used to monitor and control real-world assets remotely and, depending on the type of digital twin, perform autonomous actions. A digital twin can be thought of as a virtual representation of a physical asset. For example, a car’s digital twin would include information about the car’s characteristics, conditions, and performance. A digital twin of a building may contain information about the building’s structural integrity and energy consumption. A digital twin is a computer simulation of a real-life object or process.

Why You Need a Digital Twin

A digital twin can be used to model a process or an asset to accurately predict future behaviour or performance. This can improve decision-making and planning by providing information about the asset’s or process’ current state. Digital twins can help you improve productivity, maintenance, and safety. They can improve collaboration by providing access to relevant data from anywhere, anytime. Digital twins can also increase trust by providing greater transparency and visibility into the asset. They can also reduce costs by enabling economies of scale and reducing the need for manual verification and auditing. Digital twins are key to the future of AI and machine learning. Digital twins can be used to train AI systems and to create new products and services. Digital twins are one of the most effective ways to bring AI to the enterprise. With a digital twin, you can create a tabula rasa environment free of bias, where AI can learn and grow without being influenced by human biases.

Types of Digital Twins

Physical Digital Twin – This is a digital representation of a product’s physical characteristics and design. It can provide information about a product’s design, origin, and transformation process. A physical digital twin is useful for supply chain management and quality control. Physical Digital Twin can be used to create Physical Digital Transformation (PDT) – a process of digitizing the supply chain, where raw materials are turned into finished products. Physical Digital Twin can also be used for parts management and maintenance. Business Digital Twin – This type of digital twin is business-oriented and includes the company’s structure and processes. It can be used to simulate business processes and see how they affect customers, suppliers, and other business partners. Business Digital Twin can be used to create a Business Digital Transformation (BDT) – a process of digitizing business processes and making them collaborative and autonomous. Business Digital Twin can also be used to create Business Process Automation (BPA). People Digital Twin – Features and information about employees. It can be used to simulate working and learning processes, identify potential problems in daily activities, and create customized coaching solutions. Person Digital Twin can be used to create People Digital Transformation (PDT) – a process of digitizing employee workflows, work instructions, and training materials.

Digital Twin Benefits

A digital twin offers many benefits to companies and customers. It creates better decision-making and planning by providing information about the asset’s or process’ current state. A digital twin can help improve productivity, maintenance, and safety. It can also help increase collaboration by providing access to relevant data from anywhere, anytime. Additionally, a digital twin can help increase trust by providing greater transparency and visibility into the asset or process. Finally, a digital twin can help reduce costs by enabling economies of scale and reducing the need for manual verification and auditing. A digital twin is key to the future of AI and machine learning. It can be used to train AI systems and to create new products and services.

How to Build a Digital Twin?

There are three main steps to building a digital twin: Define the asset that will be digitized – determine what assets you want to digitize. Build a virtual representation of the asset – define the assets’ purpose, functionality, and assets’ interrelationships. Connect the asset to other systems – link the asset’s virtual representation to other systems.

Conclusion

A digital twin is a virtual representation of a real-world asset. A digital twin can be thought of as a computer simulation of a real-life object or process. It offers many benefits to companies and customers and is key to the future of AI and machine learning. Digital twins are becoming increasingly important as AI and machine learning become more widespread.

I recently read this book – Radically Human, by Paul R Daugherty and H James Wilson.

The authors are with Accenture and had written a book : Human + Machine: Reimagining Work in the age of AI, where they had discussed various case studies of how many organizations have been using AI, in the area of Human Machine collaboration.

In this book the authors propose five approaches – that they call as IDEAS as the acronym, that can be used to significantly improve the success rates of AI initiatives taken by organizations.

The key message is reassuring that AI is not to replace humans.

They talk about the first stage when the applications of AI was primarily to make routine and repetitive tasks automated. This led to discussions on whether AI will replace humans from their jobs.

The second stage was when AI approaches were used to augment or supplement human capabilities. This meant that humans could be much more accurate and much more productive. This calls for a more collaborative approach between humans and AI based solutions.

In order to achieve the benefits of this approach, one needs to relook at the five elements of innovation and by using many examples from their consulting experience with organizations across the world, the authors clearly demonstrate the adoption of these revised – possibly inverted – approaches to these five aspects.

Taken together, the IDEAS framework can be used for driving innovation across organizations.

IDEAs stands for: Intelligence, Data, Expertise, Architecture and Strategy.

While one should read the book to get a deeper dive into these factors, I will just summarize these for a quick reference:

Intelligence:

To make AI more meaningful and humanlike, one must think beyond the traditional deep learning approaches. Machines need to think and respond like humans. For example, Emotional AI is now finding applications in areas such as helping children in the autistic spectrum or in the autonomous automobiles. Two aspects they highlight are awareness and adaptability.

Data:

In order to address the struggle between the data scientists and the developers to have access to adequate data, for training or testing, approaches such as data echoing [to reuse data], active learning or synthetic data would be very useful

Expertise:

The shift that the authors propose is from ‘machine learning’ to ‘machine teaching’, where the humans can become more active teachers to help the machines learn. This will help understand a lot more context and the machine to become more intelligent!

Architecture:

The adoption of AI approaches widely has also led to the use of IoT types of devices that can constantly sense and generate a large volume of signals. For a system to be able to continuously learn and adapt to newer and emerging situations, it needs to be adaptive and scalable. This has an impact on the overall architecture of the systems. Adopting cloud based [if not cloud native] architectures and a distributed model of the systems [including microservices] to develop larger more complex solutions based on smaller components will help in making the architecture ready for the emerging future.

Strategy:

Ai technologies can significantly improve the friction usually found between strategy exercises and actual execution. With shorter cycles and closed feedback loops, strategies may be refined and fine tuned based on the execution experience. The authors call out three specific approaches that are effective for strategy. These are Forever Beta, Minimum Viable Idea and Co-lab approaches.

The book is divided into two parts.

The first discusses the IDEAS elements in detail and many examples under the theme of transforming innovation.

The second part is about competing in the radically human future, under four sub themes of Talent, Trust, Experiences and Sustainability.

The style of narration makes it a very readable book and I could learn a lot about the work already being done or that has been going on in the recent times, across the globe.

Overall, a highly recommended book for anyone wanting to get a quick update on the state of practice on AI adoption in enterprises.