Serendipity35

After writing here about how the Internet and websites are not forever, I started looking at some old posts that perhaps should be deleted or updated. With 2200+ posts here since 2006, that seems like an overwhelming and unprofitable use of my time. Plus, maybe an old post has some historical value. But I do know that there are older posts that have links to things that just don't exist on the Internet anymore.

The last post I wrote here labeled "Internet of Things" (IoT) was in June 2021. IoT was on Gartner's trends list in 2012, and my first post about IoT here was in 2009, so I thought an new update was due.

When I wrote about this in 2014, there were around 10 billion connected devices. In 2024, the number has increased to over 30 billion devices, ranging from smart home gadgets (e.g., thermostats, speakers) to industrial machines and healthcare devices. Platforms like Amazon Alexa, Google Home, and Apple HomeKit provide hubs for connecting and controlling a range of IoT devices.

The past 10 years have seen the IoT landscape evolve from a collection of isolated devices to a more integrated, intelligent, and secure ecosystem. Advancements in connectivity, AI, edge computing, security, and standardization have made IoT more powerful, reliable, and accessible, with applications transforming industries, enhancing daily life, and reshaping how we interact with technology. The number of connected devices has skyrocketed, with billions of IoT devices now in use worldwide. This widespread connectivity has enabled smarter homes, cities, and industries.

IoT devices have become more user-friendly and accessible, with smart speakers, wearables, and home automation systems becoming commonplace in households. If you have a washing machine or dryer that reminds you via an app about its cycles. or a thermostat that knows when you are in rooms or on vacation, then IoT is in your home, whether you use that term or not.

Surveying the topic online turned up a good number of things that have pushed IoT forward or that IoT has pushed forward. Most recently, I would say that the 3 big things that have pushed IoT forward are 5G and advanced connectivity, the rise of edge computing, and AI and machine learning integration:

Technological improvements, such as the rollout of 5G networks, have greatly increased the speed and reliability of IoT connections. This has allowed for real-time data processing and more efficient communication between devices.

Many IoT devices now incorporate edge computing and AI to process data locally, reducing the reliance on cloud-based servers. This allows faster decision-making, less latency, and improved security by limiting the amount of data transmitted. IoT devices have increasingly incorporated AI and machine learning for predictive analytics and automation. This shift has allowed for smarter decision-making and automation in various industries, such as manufacturing (predictive maintenance), healthcare (patient monitoring), and agriculture (smart farming).

The integration of big data and advanced analytics has enabled more sophisticated insights from IoT data. This has led to better decision-making, predictive maintenance, and personalized user experiences.

One reason why I have heard less about IoT (and written less about it) is that it has expanded beyond consumer devices to industrial applications. I discovered a new term - Industrial Internet of Things (IIoT) that includes smart manufacturing, agriculture, healthcare, and transportation, improving efficiency and productivity.

There are also concerns that have emerged. As IoT devices proliferate, so have concerns about security. Advances in cybersecurity measures have been implemented to protect data and ensure the privacy of users. The IoT security landscape has seen new protocols and encryption standards being developed to protect against vulnerabilities, with an emphasis on device authentication and secure communication.

The rollout of 5G has enhanced IoT capabilities by providing faster, more reliable connections. This has enabled more efficient real-time data processing for smart cities, autonomous vehicles, and industrial IoT applications, which can now operate at a larger scale and with lower latency.

IoT devices are now able to use machine learning and AI to learn from user behavior and improve their performance. For example, smart thermostats can learn a household’s schedule and adjust settings automatically, while security cameras can differentiate between human and non-human motion.

Edge computing has allowed IoT devices to process data locally rather than relying solely on cloud-based servers. This reduces latency and bandwidth usage, making it especially beneficial for time-sensitive applications like healthcare monitoring, industrial automation, and smart grids.

Despite the growth, the IoT market faces challenges such as chipset supply constraints, economic uncertainties, and geopolitical conflicts

 

 

 

AI agents are something of concern for OpenAI, Google, and any other players. "AI agents" are software programs designed to perform specific tasks or solve problems by using artificial intelligence techniques. These agents can work autonomously or with minimal human intervention, and they're capable of learning from data, making decisions, and adapting to new situations.

Gartner suggests that agentic AI is the most important strategic technology for 2025 and beyond. The tech analyst predicts that, by 2028, at least 15% of day-to-day work decisions will be taken autonomously through agentic AI, up from 0% in 2024. Does that excite or frighten you?

They can automate processes, analyze data, and interact with users or other systems to achieve specific goals. You probably already interact with them in applications (Siri or Alexa), customer service chatbots, and recommendation systems (Netflix or Amazon). They may be less obvious to you when using an autonomous vehicle or a financial trading system.

There are many categories into which we might place these agents because there are different types of AI agents, each with unique capabilities and purposes:

Here are some possible categorizations:

Reactive agents respond to specific stimuli and do not have a memory of past events. They work well in environments with clear, predictable rules.

Model-based agents have a memory and can learn from past experiences. They use this knowledge to predict future events and make decisions.

Goal-based agents are designed to achieve specific goals. They use planning and reasoning techniques to determine the best actions to take to reach their objectives.

Utility-based agents consider multiple factors and choose actions that maximize their overall utility or benefit. They can balance competing goals and make trade-offs.

Learning agents can improve their performance over time by learning from their experiences. They use techniques like machine learning to adapt to new situations and improve their decision-making abilities.You could also categorize agents in other ways, for example, in an educational contex.

For personalized learning, agents can adapt educational content to meet individual students' needs, learning styles, and pace. By analyzing data on students' performance and preferences, AI can recommend personalized learning paths and resources. In a related way, intelligent tutoring systems can provide one-on-one tutoring by offering explanations, feedback, and hints the way that a human tutor might. They might even be able to create more inclusive learning environments by providing tools like speech-to-text, text-to-speech, and translation services, ensuring that all students have access to educational content. By analyzing students' performance data, they could identify at-risk students and provide early interventions to help them succeed.

AI agents can automate administrative tasks for faculty, such as grading, attendance tracking, and scheduling, freeing up educators' time to focus more on teaching and interacting with students.

Agents can "assist" in creating educational materials. I would hope faculty would be closely monitoring AI creation of tests, quizzes, lesson plans, and interactive simulations.

Though I see predictions of fully AI-powered virtual classrooms that can facilitate remote learning, I believe this is the most distant application - and probably the one that most makes faculty apprehensive.

The biggest EdTech stories of 2024 seemed to all revolve around the widespread adoption of artificial intelligence.

This use of AI in education can range from fears of students using AI to "cheat," to AI-powered personalized learning. Platforms using AI to tailor learning paths to individual student needs, and providing real-time feedback and then adapting content based on progress is an AI path that most educators would welcome.

I also saw some increasing interest in skills-based training and upskilling platforms.

A positive trend is the increased accessibility to education through immersive technologies like VR/AR. I also saw positive potential with platforms addressing mental health and wellbeing within the learning environment.

Students of all ages and levels continue to acquire formal AI skills & training online: Students and workers might say they use AI in their work, but it is less likely that a person is certified in AI use in some way. (More on that in my next post.).

I saw a mention of "blended learning" in an article that reminded me of that approach that I once taught and endorsed to faculty. I have not heard the term used much in the past few years, but I am no longer involved full-time in pedagogy.

Blended learning is a pedagogical model integrating traditional face-to-face classroom instruction with online learning experiences. In some ways it was a transitional model going back to the shift from 20th century to 21st century learning. As traditional faculty were being asked to use more online tools or even convert their courses to being fully online, this approach was a softer way to launch.

The idea was to combine the best aspects of in-person and digital education to create a more flexible and personalized learning environment. A Personalized Learning Environment (PLE) was another term that emerged at the time. Probably everyone in and out of academia now has a personalized learning environment of a kind, though it may not be formalized. A PLE was supposed to allow students to benefit from direct interaction with teachers and peers while also taking advantage of the accessibility and resources available through digital platforms.

Key components of blended learning include:

    In-person instruction: Traditional classroom teaching where students engage with teachers and classmates in real-time. 
    Online learning: Use of digital tools and resources, such as videos, interactive activities, and online assessments, that students can access at their own pace.
    Integrated learning activities: Assignments and projects that blend both in-person and online elements to enhance understanding and engagement.
    Flexible pacing: Students can often progress through material at their own speed, allowing for personalized learning experiences.

Some of the advantages associated with blended learning were to provide a more dynamic and adaptive educational experience and addressing diverse learning styles and needs.

An effective blended learning module has a good range of learning activities: on-campus activities, such as lectures, workshops and seminars; off-campus activities, such as field trips, exhibitions, and visits to companies; online synchronous activities; and independent learning activities, such as completing tasks after reading case studies or watching videos.

The article I read was from the UK timeshighereducation.com and had suggested goals for blended learning. In brief, they are:
Find a suitable space when attending online classes
Use digital tools
Create a sense of belonging (a difficult goal because online interactions often feel impersonal and might not be well suited to every student - or faculty member)

One suggestion that interested me the most was to use different types of assessment. This was an area that I worked with faculty on frequently as an instructional designer. Blended learning modules should use a good range of assessment types. It was difficult for many teachers to accept that their main form of assessment was testing, especially objective, knowledge-based tests and quizzes. Written assessments, such as reports and essays, appeared in some courses (especially in the humanities) but were often absent in STEM courses. Faculty would tell me, "They are too subjective." "They take too long to grade" "My course requires them to retain lots of facts that I have to assess." The latter was especially true in foundation courses.

Using online tests and quizzes became more popular because once created they could be automatically scored. Easy for the teacher and immediate feedback for the student.

In-person or recorded presentations were more in the blended model but were time-consuming and more popular in upper-level or graduate courses. Interacting face-to-face with their peers as a team or audience during the presentation is also an important skill. I saw video presentations, e-portfolios, digital projects, posters, podcasts and simulation games all used in blended courses. 

One concept that often met with faculty indifference or opposition was the student-as-co-creator of assignments and assessments, though this can serve as a valuable source to gather student voices and improve their learning experience.

The term "hybrid course" became used more than "blended" but was often the same thing or just used interchangeably. While both models integrate online and offline learning, blended learning is a broader pedagogical approach that can be applied at various levels of education and in different ways. A hybrid course is a specific type of course design commonly used in colleges and universities.

A hybrid course refers to a course that "officially" combines face-to-face (F2F) classroom instruction with online components. The term is commonly used in higher education to describe courses where a significant portion of the learning activities are conducted online, with the remaining portion happening in a physical classroom setting. This becomes an issue concerning the registrar and scheduling areas. A course that met F2F on Tuesday and Thursday from 10:30 - noon may now only be assigned a classroom on one of those days. The goal is still to balance the in-person interaction with the flexibility of online learning, usually reducing the amount of time spent in a physical classroom compared to a traditional course.

Of course, hybrid learning models should not be used simply to free up classroom space or reduce parking issues on campus, but unfortunately, I knew of cases where that was a motivation for using it.

The development of online and blended learning modules got a boost during and after the pandemic. To a degree, that was from necessity and convenience, but it introduced these approaches to more students and more faculty and some of it has remained in use.

Developing the right balance between these different teaching modes varies according to discipline, but a mix of synchronous (real-time) and asynchronous (self-paced) online activities, along with in-person classroom sessions.is still the pedagogical approach.