Humans and AI are two sides of the same coin

We, at Excelsoft, have been pioneers in adopting advanced technologies into our products at an early stage of their inception and keep pace with the evolution of the technology. In the same vein, we see immense potential in Artificial Intelligence(AI) to transform the Edtech landscape by providing new experiences. We deeply think about meaningful adoption of AI into our products.

How AI can help in Human evolution

One of the key aspects in adoption of AI is to understand utility of AI is dependent on the fundamentals humans have in the sphere of activity -say learning. Therefore, there is a necessity for AI assistants to cater to the knowledge levels of humans. Then evolve the human understanding to higher states. As an example, we will see the use of AI in the sport of chess. Chess players were pioneers in using AI as an assistant. This led to an advancement of knowledge of chess concepts and patterns. AI engines such as Stockfish changed the game forever. A single chess match used to happen over days with many adjournments of a position. The teams had to  work out manually on the right moves their player must play on resumption. The evolution of human understanding of chess, driven by AI engines, has led to matches now being challenged within 7-8 hours.

Today’s AI needs to be adapted to human comprehension

In due course of time, AI engines became more powerful after training on higher quality games of evolved players. At the stroke of the millennium, they were evaluated against very strong chess players such as Kasparov and Kramnik. The realization was humans cannot defeat the brute power of AI chess engines. In the following two decades, they have become contextually aware and more powerful that the World No. 1 Carlsen has said he does not fancy playing against AI engines. These events elucidate the fact human intelligence trained over several decades are never a match to AI trained over a few months or days. Understanding AI’s output requires a sufficient level of human development. So, it is essential to calibrate AI to aid within the realms of human comprehension. Funnily, the all-powerful AI’s utility is constrained by what humans can or cannot understand. 

How human evolution can make AI stronger

It also highlights another important fact. Advancement of human understanding has led to more powerful AI engines in chess as the quality of training data improved over time. On the one hand, there is a huge amount of investment being made in the field of AI to aid human lives. Lest we forget that AI has no traceability, and their usage needs to be understood by humans uncovering the patterns in the responses. Understanding that humans and AI need to co-exist to sustain AI is essential. AI is a developing field, and practitioners are still beginning to understand its potential uses. Therefore, to ensure useful progress of AI continues, we must actively discuss the future of AI in its early stages. We need to drive and invest in human evolution to unleash the full potential of AI. As the saying goes, iron sharpens iron, AI and humans must sharpen each other. This human and AI harmony must be there to ensure AI is a force of good.

Let us take a use case in learning to explain this. For example, there are understandable constraints for teachers to follow a personalized teaching approach. Let us take an example of a math teacher in Class XI teaching a student how to find slope of geometric figures. Since eleventh grade curriculum dictates use of calculus to do the same the teacher may not adopt a different approach towards a person still trying to understand finding slopes using more straightforward methods. In this scenario, an AI based teaching assistant can understand the current knowledge level and follow an approach to evolve the student from simpler to advanced concepts. Ultimately, this student will start scoring well, quality of data generated is high-quality which makes the AI engine stronger and progress to the next level.

Read these blogs to discover the latest insights on AI in Education :

Exploring the Role of Large Language Models in Education

I first came across a complete AI tool on November 30, 2022, when I learned about ChatGPT. Initially, I dismissed and shunned it. 

Sometime in Jan 2023, at an ID workshop, ChatGPT was the main attraction, and almost everyone saw it as an adversary, and the atmosphere of rejection was in the room. I decided not to use it as my peers had the same opinion. But there is a saying in English, “Man proposes and God disposes!”

By March 2023, I understood the mistake when I gave it a try for the first time. It was so helpful that I couldn’t resist using it.

Today, at 53, it is like my junior, the smartest one, who never takes a break, sleeps with me, and gets up at the same time. The day is not far when it will work while I sleep.

So, with ChatGPT, “Do I matter?” 

Good, bad, or ugly, this is the truth, whether we like it or not. We need to upgrade our skills to keep it as the best assistant. 

In the last nine months, I have witnessed rapid change, and my aim is to adopt the practices that will help me cope with this change.

In the recent years, AI has transformed almost all the professions. According to the LinkedIn 2023 “Future of Work” report, from December 2022 to September 2023, discussions on AI have increased by 70%. The report also stated a change in skill sets for almost all the job roles by 2030.

Last year, many of my friends were afraid of losing their significance, their work, and their position in the world. But now, they are more optimistic and self-assured. They slowly realize that AI tools are not a danger but a help to their work. AI tools can enhance the quality and productivity of their work by incorporating AI components into their tasks.

Let’s embrace the AI tools with an open mind and make them serve you. They are not your competitors; they are the facilitators of innovation. It is just changing our conventional environment… our home, social places, work, and many more. But it is not altering the structure of our life! 

We just need to get ready for some major changes. 

Change is unavoidable, and the people should be ready to face it. 

Now, let’s examine the effects that tools like ChatGPT can have on education. Education as an industry is undergoing a fast change, and for the last two decades, it has been accelerating like a meteor. What intrigues me the most is the persistence of the thought leaders trying to improve learning during this period. Finally, we got a chance to see ChatGPT, which can disrupt the way students learn. 

Actually, learn without limits!

Pandora’s Box is open now! Are we anticipating troubles?

Similar to other industries, people had the same opinion at first, but the view has shifted, and people in the educational sector have embraced it fully. ChatGPT is one of the earliest LLMs to enter the market. It is the most popular topic, and these LLMs are well suited to offer learners limitless learning opportunities.

LLMs have the power to sketch our future.  

LLMs are gaining popularity and gaining acceptance all over the world. But let me share the two incidents that made me write my doubts about its usage. 

The first one is about my daughter, a tenth grader. I have always been keen to try out new technological progress. So when she started using a well-known tool, I was very pleased as I thought questing could lead to better learning, and this tool exactly does that.

However, last week, I saw her asking for a tool to create a question paper for class X. I was shocked that ethical behavior was lacking. When I questioned her, she explained that others are doing the same and she should not be left out. I let her continue the search for the elusive exam paper.

God, have some pity the results that were shown were quite comprehensive; with few changes in subject and exam features, she finally got an exam paper.

After a week, when she took her preparatory exam, I was amazed to see an 85% match. That means 100% of students will get 85% on their exams, and that’s the new zero. This is inflation of some sort, which is here to stay.

But what about the hollow minds in the process?

The second is the statement of the UN chief endorsing the idea of a global AI watchdog like the International Atomic Energy Agency (IAEA). Guterres told reporters, “The latest form of artificial intelligence – generative AI – is ringing loud alarm bells…“We have to heed those warnings.” He has revealed plans to initiate work on a high-level AI advisory body to regularly assess AI governance arrangements and offer suggestions on how they can conform to human rights, the rule of law, and the common good.

People support the idea of Open AI advocates forming a body like the IAEA that could impose restrictions on deployment, check compliance with safety standards, and monitor the usage of computing power. British Prime Minister Rishi Sunak has also backed the idea and wants Britain to be the leader of global AI safety regulations.

So, there is no peace on the Western front!

People have understood the power of AI tools, and providing equal and regulated access for everyone is the key. As we all know, “Equal opportunity” and “Democratization” are two ways of distributing power. Till any consistent resolution is adopted the world over, we need to do the moderation and propose a global framework for Ethical AI Use in Education.

So, how is it impacting education?

The advancement in AI technology, especially LLMs like ChatGPT, has changed many aspects of our lives, including education. While these tools provide access to information and ease of use, they also create difficulties regarding the originality of student work, academic honesty, and accurate evaluation of learning. 

The education industry needs to make a quick decision on the ethical use of AI before it becomes too late. Moreover, the ease of use has resulted in over-dependence, raising various concerns from educators such as: 

  • Is work produced with the significant help of LLMs can be considered original?
  • Are students learning well with LLMs?     
  • How can institutions guarantee the integrity of academic work?

All these concerns require a systematic approach and a Global framework in educational settings. A framework maintains a balance between using the benefits of LLMs and upholding academic standards. Such a framework will ensure fairness in student’s academic endeavors, ensuring equal and just opportunity. In addition, the framework should be adaptable, having clear guidelines for efficient supervision by any educational institution.

The integration of AI tools in education is inevitable and beneficial, but it requires careful management to preserve the integrity of academic work. I am inviting all academics worldwide to contribute toward this universal framework. 

Our brilliant minds at Excelsoft Technologies have proposed a set of well-researched guidelines for our partner schools on LLM usage:

  • Specify an LLM for students on school premises. 
  • Use LLMs as an additional Tool, and everyone after the eighth standard will have proctored access.
  • Use of LLMs is only allowed for students in pairs using one machine with a session proctoring feature.  
  • Use LLMs for initial research and not as a source of content and add a reference in the final write-up.
  • Use LLMs for idea generation and discuss it with your teacher for extrapolating it. 
  • Any use of LLMs for academic work will be disclosed by the students and teachers.
  • Any LLM-generated content must be modified by the student before submission.
  • Students should be prohibited from submitting LLM-generated content as their original work.
  • Use LLM for the generation of assessment items.
  • Use LLM for Feedback augmentation.

We help educational institutions formulate their academic policies along with disciplinary actions for any non-compliance. In addition, we have set up resources and mechanisms to monitor and enforce these guidelines. We conduct mandated periodic reviews to accommodate technological and academic advancements.

Remember, AI tools will constitute tomorrow’s workforce; they will run the cities in the future, and every industry will involve AI. They allow learning without limits.

Let’s embrace it with Caution!

At Excelsoft, I am excited as I work on LLM tool integration that can offer

  • Integrity Protection
  • Learning Assistance
  • Skill Augmentation
  • Bibliographic Reliability
  • Enhanced Readability with Improved Focus
  • Language Refinement and Vocabulary enrichment

Read these blogs to discover the latest insights on AI in Assessments:

For more information on our pursuit, reach out to us at connect@excelsoftcorp.com

Explore this and many such interesting articles at the eLearning Industry

Unveiling the Hidden Biases: A Deep Dive into Bias in Artificial Intelligence

As AI rapidly transforms various aspects of our lives, it is imperative to address the latent biases that can permeate AI systems. This article delves into the nuances of bias in AI, exploring its sources, implications, and potential solutions to mitigate its impact.

Sources of Bias in AI:

1. Data Bias:

The crux of bias in AI often lies in the data used to train AI models. If the data is biased, either intentionally or unintentionally, it can lead to biased outcomes. For example, training data that lacks diversity in terms of gender, race, or socioeconomic status can perpetuate existing biases in AI systems.

2. Algorithmic Bias:

AI algorithms, designed by humans, can inherently harbour biases. Hard-coded assumptions or preferences can lead to biased decision-making. For instance, an algorithm designed to predict recidivism rates may be biased against certain demographics due to historical data that overrepresents particular groups.

3. Human Bias:

As AI systems are developed and deployed by humans, they can inadvertently inherit biases from their creators. Conscious or unconscious prejudices, cultural norms, or personal experiences can influence the design, implementation, and evaluation of AI systems, leading to bias.

Implications of Bias in AI:

1. Unfairness and Discrimination:

Bias in AI can result in unfair and discriminatory outcomes, particularly for marginalised groups. For example, biased AI systems in hiring algorithms may lead to discrimination against certain demographic groups, limiting opportunities for employment.

2. Lack of Trust and Accountability:

When AI systems exhibit bias, trust in their outcomes can erode. This lack of trust can hinder the adoption and utilisation of AI technologies, diminishing their potential benefits. Additionally, it can be challenging to establish accountability for biased outcomes, as the complexity of AI systems often makes it difficult to trace the origin of bias.

3. Ethical and Societal Concerns:

Bias in AI raises profound ethical and societal concerns. Unchecked biases can perpetuate systemic inequalities, exacerbate social divisions, and undermine fundamental human rights. The implications extend beyond individual cases of bias, potentially affecting entire communities and shaping societal structures.

Mitigating Bias in AI:

1. Data Curation and Preprocessing:

Addressing bias in AI begins with the data used to train models. Techniques such as data augmentation, resampling, and algorithmic fairness can help reduce bias in training data, leading to more equitable outcomes.

2. Algorithmic Auditing and Fairness Evaluation:

Regularly auditing AI algorithms for bias is crucial. Fairness metrics and evaluation frameworks can help identify and quantify bias in AI systems, enabling developers to take corrective measures.

3. Human Oversight and Regulation:

Human oversight and regulation play a vital role in mitigating bias in AI. Establishing clear guidelines, standards, and accountability mechanisms can help ensure that AI systems are designed and deployed responsibly, minimising the potential for bias.

4. Education and Awareness:

Promoting education and awareness about bias in AI is essential. AI professionals, policymakers, and the general public must understand the sources and implications of bias to drive collective action towards more ethical and inclusive AI development.

Conclusion:

Bias in AI is a multifaceted issue that requires a multidisciplinary approach to address. By recognising the sources and implications of bias, stakeholders can work together to develop responsible AI practises, ensuring that AI technologies benefit all members of society fairly and equitably. The journey towards unbiased AI is an ongoing process that demands continuous vigilance, collaboration, and commitment to ethical AI development.

AI in Education Technology

At Excelsoft, we have recognised the need to take up an organisation-wide initiative to provide thrust and impetus to artificial intelligence and other related technologies. We have not only been emphasising the need for an ‘urgent AI push’ but have also been working in a clear direction in this regard.

There have been a variety of activities and initiatives that are already underway across the organisation. AI-driven features and functionality have been introduced into our products. There have been several proofs-of-concept (POC) of using AI tools and technologies. Our sales team has been positioning these products and solutions for our customers. We are conducting workshops and seminars frequently to keep all teams up-to-date on the latest developments in AI. Many departments have incorporated AI tools into their functions. Many of our colleagues have been using AI in their day-to-day work.

While all these actions are encouraging, we realised that there is a strong need to formulate a clear plan for an ‘organisation-wide AI strategy’ and execute the same in a structured manner as a programme. Earlier this year, we created an AI Focus Group at Excelsoft that will take ownership of this programme.

AI Focus Group at Excelsoft

The AI Focus Group focuses on the following core objectives:

  • Organisation Level Guidelines: For use of AI and ML in Products and Services
  • Alignment of AI Initiatives: Across all Products and Services
  • Capability Building: For Product, Tech and Operations Teams
  • Standard Tools and Libraries: Validated set of Tools and Libraries for Product, Tech and Operations Teams
  • Prototypes and Features: For Customer and Market-Driven Use-Cases
  • Legal and Ethical Considerations: For use of AI in Products and Services

AI in Processes and Operations

We have started incorporating AI into our processes and operations. Below are some key areas where we have been successful:

  • Development: AI-Assisted Code Generation, Review, and Correction has clearly increased the productivity of our developers across technologies. We are working closely with Microsoft and Amazon, exploring their tools in this effort.
  • Quality Assurance: AI-Assisted Generation of Test Scripts, Test Plans, and Test Cases has increased the overall quality and efficiency of our QA Teams.
  • Deployment: AI-Assisted Automated Deployments and Issue Prediction are being experimented with for seamless, issue-free deployments.
  • Technical Documentation: AI-Assisted Specifications Documentation and user manuals are being tried out for quick, accurate technical documentation.
  • HR and Learning and Development: The HR teams are actively using AI tools for screening CVs and AI-Assisted Policy Documentation.

AI Features in Products

Over the past few months, we have worked with our customers and gathered insights from forums and industry experts to identify key market needs that can be solved by applying AI. Below are some active projects we have been undertaking:

Online Assessments:

  • AI-Assisted Item and Test Construction
  • AI-Assisted Auto-Marking of Written, Spoken and Programming Responses
  • Smart Post-test Forensics
  • Smart Flagging of Duplicate and Compromised Items

Remote Proctoring:

  • AI in Remote Proctoring: Face, Voice, Object, Keystrokes Detection
  • Selective blurring of test content for proctors

Learning Experience and Content Services:

  • AI-Assisted Content Generation and Beautification
  • AI chatbots for Content Discovery and Tutoring
  • AI for Student Profiling and Personalised Learning
  • AI-Powered Content Recommendations

Student Success and Higher Education:

  • Predictive Analytics and Early Alerts
  • Detection of At-risk Students

Technologies and Models

Excelsoft has been an early adopter of the OpenAI platform and APIs. We have extensively used ChatGPT 3.5, GPT 4.0, and DALL 3.0 in our prototypes and features. In addition to OpenAI, we have explored other well-known LLMs:

  • Llama 2 (from Meta)
  • Bard (from Google)
  • Stable Diffusion (from Stability.ai)
  • Titan (from Amazon)

Risks and Challenges

While AI has in general shown great promise in various applications, we are also fully aware of the potential risks and concerns associated with its use, including bias, fairness, misinformation, fake content, ethical concerns, security threats, and legal and privacy issues.

We continue to observe this space closely. We participate in several AI communities and are working on setting up strong guidance principles for the use of AI in our products and services. We have also taken several measures to prevent any violation of the legal or ethical considerations of our employees and customers.

Moving forward, we are committed to executing our organisation-wide AI strategy in a structured manner.

The Age of AI in Education: Promises and Concerns

Remember AlphaZero that defeated Stockfish, which was the best chess program in the world? In 2017, it won 28 games, drew 72, and lost none. The next year, while playing 1000 games against Stockfish, it won 155, lost 6, and the rest were held a draw.

So what was so special about AlphaZero?

AlphaZero had no predefined moves or strategies from human play. It was a genuine product of Artificial Intelligence (AI) training where developers keyed in the chess rules and coded the program to develop strategies that improve the chances of winning every match. Just a self-training of 4 hours made AlphaZero the world’s most effective chess module. No human has ever beaten it.

It’s time we take AI and Machine Learning (ML) seriously. They are showing immense potential to create new types of businesses across business operations with striking results.

To check the popularity, I immediately started my research with Google. Searching “What is Human intelligence?” fetched 10 billion results in 0.42 seconds, whereas “What is Artificial Intelligence?” generated about 5.8 billion results in just 0.52 seconds.

The day is not far when AI (Inorganic Intelligence) will leave Human intelligence(Organic Intelligence) behind.

Welcome to the new world, where organic is meeting the inorganic.

AI is making that happen everywhere, and it’s viral!

AI is not an industry or a domain, let alone a single product. It is an enabler that has the capacity to learn, evolve, and surprise. It will disrupt and transform the human experience to levels that were not experienced up until now.

So, can we call this a new age?

As you know, human civilizations progressed with Stone Age, Bronze Age, and Iron Age, developing competencies using the materials and technology of each of the ages. Till the end of the 20th century, humans with better cognitive abilities have enjoyed success at work. However, with computing advancements in the 21st century, we are experiencing a disruption that has embarked us on a new age of freedom, where information and decision come for free.

We live in the AGE of AI

Every day, everywhere, AI is gaining popularity, and one of the biggest beneficiaries might be education. While AI has the potential to revolutionize the way we think about education, there are still many challenges and concerns that need to be addressed.

By the way, the presence of AI in Education is significant, with a valuation of $4 billion in 2022, and is projected to expand over 10% CAGR from 2023 to 2032. AI in Education Market Statistics, Trends & Growth Opportunity 2032 (gminsights.com).

In our recent Townhall, almost every conversation on Education Technology converged to AI and its impact. A question that gained everyone’s attention was, “In Education, who should benefit the most from AI?

Our CEO’s response was very candid, “AI should impact the student experience and improve performance outcomes.” He also added that the other stakeholders will also get their share of the pie by analyzing the insights from student engagement.

Let me summarize the takeaway.

In the EdTech domain, AI should serve the “King” aka THE LEARNER.
When they benefit, other stakeholders will get a share of the pie.

The big question revolves around the impact that AI will bring on education.

On my way to Mumbai, I had a conversation with a principal. He was both excited and skeptical about AI and was seeking answers to his questions.

  • What does AI-enabled education look like?
  • Will AI replace human intellect or critical thinking?
  • What do AI-enabled assistants mean to children?
  • Will AI-based assessments used for shaping human action be allowed?
  • Will cheating be rampant and cripple the learning model in education?
  • Should we teach children how to frame compelling questions and AI will do the rest?

I explained to him that the integration of AI in education could provide dynamic learning environments that are accessible, engaging, effective, and offer personalized learning experiences with intelligent tutoring systems.

But I still saw the worry in his eyes. He said, “My friend, it is important to strike an equilibrium between technology and human interaction, ensuring safe and secured learning environments.”

His words made me think about the opportunities that AI can offer to students and also its ramifications. AI can:

  • Personalize learning with tailored learning interventions, real-time feedback and opportunity for graded practice.
  • Enhance learning support and prioritize learning interventions along with improved assessment quality.
  • Present immersive experiences and enhance learner participation in regulated real-world situations.
  • Increase accessibility for learners.
  • Save costs by automating difficult tasks and facilitating customized instructions.
  • Revolutionize smart content creation by generating tailored learning materials.
  • Enhance the ease of performing administrative tasks and improve the efficiency of learning delivery.
  • Provide access to educational resources, particularly for students with limited access.
  • Analyze data patterns by using AI algorithms to detect early warning signs and alert educators for timely interventions.

However, we also need to be responsible enough to address the following:

  • Strengthen the bias and Inequality algorithms with continuous monitoring. This will help remove the learning and assessment bias.
  • Address privacy and security concerns as AI-powered learning systems gather a wide range of student information, including their behavior, learning progress, and personal data.
  • Address the tech dependencies and their impact on learners’ critical thinking and problem-solving skills. Even a continuous upgrade of modern technology to ride this AI wave is worrisome.

AI has immense potential to revolutionize the education sector. However, legislatures, policymakers, entrepreneurs, educators, and technology enthusiasts must work together to ensure that AI-driven learning platforms and tools are used ethically and responsibly.

We at Excelsoft Technologies have a dedicated R&D team that has undertaken a keen interest in formalizing and expanding the use of AI technology and has been working on the following pointers:

  • Define policies and guidelines to guide the use of AI for all the stakeholders.
  • Publish ethical implications of using AI in education, including data privacy, security, misuse, and algorithmic bias concerns.
  • Research the latest trends and developments in education technology, including current applications, SAAS business models, and Integration APIs for potential future use cases.
  • Present the out look of different stakeholders in evaluating the potential benefits and risks of AI in education.
  • Develop specific manifestations of how AI is used in our demonstration school (Excel Public School) and evaluate its effectiveness.
  • Formalize the training and development needs of the stakeholders and ensure the effective incorporation of AI in the education system.
  • Define the process of monitoring and evaluating the ethical usage of AI.

In conclusion, AI holds great promise for the future and will play a pivotal role in shaping our lives. The integration of AI in education holds tremendous potential for transforming learning environments. By leveraging AI’s capabilities, education can become more accessible, engaging, and effective.

We at Excelsoft focus on the following to achieve good outcomes in the field of AI:

  • Ease of deployment with integration APIs and process data analytics
  • Integrating skills and competencies with strong learning analytics and reporting
  • Performance management with adaptive testing and evaluation
  • Promote responsible actions with state-of-the-art proctoring and evidence-based practices

For more information on our pursuit, reach out to us at connect@excelsoftcorp.com