The article explores the development of bots from early automated systems to advanced AI agents, highlighting both their potential and limitations. It also discusses the future of AI, emphasizing the need to address societal inequalities as AI technology advances.
Bots have been used to communicate with humans for a few decades, starting as early as the 1980s with Interactive Voice Response (IVR) systems. These systems are typically used to answer customer calls, like those used by banks where you say “yes” or “representative” to be automatically connected (Pawlewicz, n.d.). They have also been used to communicate with sensors or data. LabBot is an excellent example; it uses a Telegram bot that can be programmed to monitor sensor data and send automatic warning messages when the sensor exceeds limits (Alexriss, n.d.). A Telegram bot uses the Telegram messaging service app., to communicate with users allowing users to send commands and receive information, such as sensor readings or control outputs from devices (Victor, 2020). However, one of the most fascinating bot interactions is when bots interact with each other. The article “Even Good Bots Fight: The Case of Wikipedia” by Tsvetkova et al. (2017) is a study that analyzes the interactions between Wikipedia bots between 2001 and 2010. Most Wikipedia bots are designed to edit and link Wikipedia articles, the study found that the bots often end up undoing each other’s edits, sometimes leading to prolonged editing conflict between bots.
What is a Bot?
A bot is a program or a collection of programs that performs automated tasks usually over the Internet. In the context of the Wikipedia encyclopedia, bots are owned by Wikipedia editors and can perform a wide range of tasks, from fixing broken links and categorizing pages to undoing acts of vandalism (Kohli, 2018). They can be considered a form of Narrow AI. Narrow AIs are systems designed to perform specific tasks and operate under limited constraints (Awan, 2023).
Wikipedia Bots Interactions
Wikipedia Bots are programmed to perform specific tasks like editing, linking pages, or correcting errors (Kohli, 2018). Bots, in some cases, make over 50% of the edits in smaller Wikipedia language editions (Tsvetkova et al., 2017). This editing task and maintaining datasets, notably large datasets would be overwhelming for humans alone. Thus, the bots play a crucial role in maintaining and keeping the encyclopedia up-to-date and relevant. However, during the period from 2001 to 2010, Wikipedia bots were known to undo each other’s edits repeatedly and could be described as “bot wars” or bots editing wars (Godosh, 2017). These editing war-like conflicts, in some cases, went on for years wasting time and resources. This occurred because the bots were following slightly different editing rules or interpretations of how to edit articles and were not programmed to collaborate. Moreover, this highlights the challenges involved in creating logical agents that follow a rigid set of rules. Logical agents are AI agents that act based on a Knowledge Base (KB) (and a set of rules) to infer (to derive) new knowledge or make decisions accordingly (Russell & Norvig), in the case of the Wikipedia bots to or not to edit or undo edits. In other words, Wikipedia bots reflect the shortcomings of Narrow AIs which have narrow KB, lack flexibility, and are missing contextual understanding needed for deep communication and collaboration.
The Future of Bots
However, with the emergence of Large Language Models based on the Transformers neural network architecture such as ChatGPT and Claude, bots can be transformed into AI agents capable of autonomy performing more than just one narrow task but a multitude of tasks, capable of communicating, collaborating, delegating, and even hiring other AI Agents and humans to perform task for them. According to an Accenture Report (2024, p. 73), “96% of executives agree that leveraging AI agent ecosystems will be a significant opportunity for their organizations in the next three years.” Fully agentic AI is predicted to appear on mobile devices and PCs as early as 2025. This will fundamentally transform how people from the modern industrialized society learn, work, communicate, and live their daily lives, as well as impact the rest of humanity, hopefully for the better…
What I would like to see implemented.
I would like to talk to my computer and ask it to perform tasks for me, such as opening a file for me while I am writing a paper in a different file or coding. In other words, I want an AI Agentic Operating System for PCs capable of interfacing through voice, vision, and text; including mouse, keyboard, and touch screens. This would allow me to multitask more efficiently, making my life much easier.
On a more serious note, the 2024 Nobel Prize in Physics went to J. Hopfield and G. Hinton for machine learning discoveries. Hilton also called the godfather of AI, said “I am worried that the overall consequence of this might be systems more intelligent than us that eventually take control” (Pollard & Ashlander, 2024, p1). Hinton’s concerns are grounded and very real. However, my main concern, not to dismiss Hilton’s concern, is that AI will generate a greater gap between industrialized countries and developing nations exacerbating existing inequalities; and in the long run, creating a speciation situation of the human race (e.g. augmented and none-augmented human) — Between those that can afford human augmentations brought through AI and technologies such as Crisper, Anti-Aging, and Neuralink and those that can’t. Thus, what I really would like to see is the implementation of AI that benefits not just a select few, but humanity as a whole.
To summarize, bots have evolved from simple automated systems to sophisticated AI. The future promises more AI agentic systems capable of autonomously handling complex, multi-task operations, potentially transforming how we work and live, with technological and scientific advancements that would seem magical. It is up to us to ensure these advancements are used for the good of all humanity.
References:
Accenture Report (2024, January 9). Technology Vision 2024: Human by design [PDF-Slides]. Accenture. https://www.accenture.com/content/dam/accenture/final/accenture-com/document-2/Accenture-Tech-Vision-2024.pdf#zoom=40
Alexriss, (n.d.). LabBot. Github. https://github.com/alexriss/labbotLinks to an external site.
Awan, A. (2023, June 2023). What is narrow AI? Datacamp. https://www.datacamp.com/blog/what-is-narrow-ai/
Godosh, S. (2017, February 27). Wikipedia bots spent years fighting silent, tiny battles with each other. Popular Science. https://www.popsci.com/wikipedia-bots-fighting/
Kohli, N. (2018, October 16). The bots that help run Wikipedia [Video].
Strange Loop Conference — YouTube. https://www.youtube.com/watch?v=mOmLiAPdF0Y&t=50s
Pawlewicz, K. (n.d.) A history of customer service. Olark. https://blog.olark.com/the-history-of-customer-service/
Pollard, N., & Ahlander, P. (2024, October 2024). Nobel physics prize 2024 won by AI pioneers John Hopfield and Geoffrey Hinton. Reuter. https://www.reuters.com/science/hopfield-hinton-win-2024-nobel-prize-physics-2024-10-08/
Tsvetkova, M., García-Gavilanes, R., Floridi, L., & Yasseri, T. (2017). Even good bots fight: The case of Wikipedia. PLoS One, 12(2).
Russell, S. & Norvig, P. (2021). 7. Logical Agent. Artificial intelligence: A modern approach. 4th edition. Pearson Education, Inc. ISBN: 9780134610993; eISBN: 9780134671932.
Victor (2020, July 9). Telegram: Request ESP32/ESP8266 sensor readings (Arduino IDE). Random Nerd Tutorials. https://randomnerdtutorials.com/telegram-request-esp32-esp8266-nodemcu-sensor-readings/
Originally published at Alex.omegapy on Medium published by Level UP Coding on October 23, 2024.
Top comments (0)