Human: Say Hello To Your New Colleague - AI Agent (Employee)
Human: Say Hello To Your New Colleague - AI Agent (Employee)
The biggest change to personal computing in the past 50 years or so is coming. The “personal” in computer is about to become that truly.
Almost a year ago, in one of my many newsletters, I had written about this new species called "AI agents", and how they would change the way we humans lived offline and online.
Coincidentally, Bill Gates of Microsoft fame, too, had claimed something similar in his newsletter. I had then proceeded to explain what AI agents really were.
A few days ago, Microsoft announced it would launch autonomous AI agents for Copilot Studio and Dynamics 365, allowing users to create custom agents and choose from pre-made options for business deployment.
Microsoft now calls them "AI employees", but essentially, these are advanced AI agents designed to take on more sophisticated roles within organizations. These AI employees will be capable of performing tasks that require higher levels of judgment and decision-making, further integrating AI into everyday business operations.
It's begun.
Of course, while announcing the news, Microsoft added the now-standard remark that almost every other AI company makes - although some human roles may shift with the rise of AI agents, the overall impact would be positive, creating new job opportunities — much like past technological revolutions such as the printing press and the PC.
Here's the thing - I don't know which "new jobs" Microsoft is talking about (I leave it to the IT giant's wisdom for now) but I would not be exaggerating if I were to claim that AI agents are going to make a 360-degree change in our lives.At the cost of humans.
Very quickly, for those who are unaware, think of AI agents as "intelligent" or "smart" software programs living inside your PC or phone, designed to perform tasks autonomously on behalf of users or other systems.These agents can understand their environment, make decisions, and take action to achieve specific goals.
AI agents, as I love to explain, will be your personal assistant, butler, grocer, and what-may-have-you all rolled into one. At your place of work, it can be whatever you want it to be.
Just think of them as digital assistants or robots that can handle complex tasks without constant human intervention or even supervision.
Example: A doctor can use an AI agent to handle all his patient appointments, cancellations, re-schedulings and meetings by merely giving the agent access to his calendar and appointments' diary, and patients' Rolodex. At the other end, the agent will also be able to monitor a patient's medicine schedule, intake, and medicine adherence, and will even notify the doctor if anything in the routine is amiss.
So if you thought the simple AI-assisted chatbots were great, wait till you come across an AI agent.
Alert: AI agents and chatbots are not the same.AI agents are far more technologically sophisticated, and capable of handling a wider range of tasks.
Join one of the fastest-growing online communities around artificial intelligence - AI For Real.
Don't believe me? Here are just some of the key features of an AI Agent so decide for yourself:
Autonomy: AI agents can operate independently, making decisions and executing actions based on their programming and the data they receive.
Learning and Adaptation: They can learn from their experiences and improve their performance over time. This means they can adapt to new situations and refine their actions to better meet their objectives.
Task Execution: AI agents can be assigned various tasks, from simple ones like scheduling meetings to complex ones like managing inventory in a warehouse.
Interaction: These agents can interact with humans and other systems, often using natural language processing to understand and respond to user inputs.
So how different are AI agents from chatbots? In the interest of brevity, let me just give you two examples which will help you understand their superiority:
AI agents can recognize the intent and the emotions of a user/customer. So the agent can analyze the sentiment behind words or actions, identify feelings like anger or anxiety, and respond accordingly. Chatbots have a large, structured data bank and merely follow a written script by and large. Ask them an "out of context" issue and you'll how they go haywire. Will rarely happen with an AI agent, though.
But that's just the customer-facing side. You can create and use an AI agent from within your PC to help you with tasks such as appointment scheduling, re-scheduling, tasks allotments, all without your actual intervention or monitoring. It's like a "fire and forget" missile.
Companies and research labs like Adept, crewAI, and Imbue are moving from knowledge-based models (chatbots) to action-based ones (AI agents), focusing on developing agent-based models and multi-agent systems. With the rapid advancement of generative AI, these agents could soon become as common as chatbots are today.
A few companies using Copilot have reported significant productivity gains, but the use of AI agents may also lead to potential job losses as tasks are completed more efficiently. Which does not harbor well for human workers now, does it?
The Reason Behind AI Agent's Superiority
Why are AI agents superior and something that humans have to fear? That's because they can learn and improve over time through several key mechanisms. Till a few years ago, it was not possible to develop an AI agent because AI was not really mainstream (for several reasons) and because the amount of computing power required was not available, among other things.
Today, though, with access to unsupervised learning, and even better, natural language processing, machines understand and communicate with humans far better than a few years before.
Machine Learning (ML)
Supervised Learning: AI agents are trained on labeled datasets, where they learn to make predictions or decisions based on input-output pairs. For example, an AI agent might learn to classify emails as spam or not spam by being trained on a dataset of labeled emails.
Unsupervised Learning: Here, AI agents find patterns and relationships in data without labeled outcomes. This is useful for tasks like clustering customers into different segments based on their behavior.
Reinforcement Learning: AI agents learn by interacting with their environment and receiving feedback in the form of rewards or penalties. This method is often used in scenarios like game playing or robotic control, where the agent learns the best actions to take to maximize its rewards.
Natural Language Processing (NLP)
AI agents use NLP to understand and generate human language. They improve their language skills by training on vast amounts of text data, learning to recognize context, sentiment, and intent. This allows them to interact more naturally with users.
Continuous Learning
AI agents can continuously learn from new data and experiences. For instance, a customer service AI agent can improve its responses by analyzing interactions and feedback from users, adapting to new queries, and improving its accuracy over time.
Feedback Loops
User feedback is crucial for AI agents. Positive feedback reinforces correct actions, while negative feedback helps the agent learn from mistakes. This iterative process helps the agent refine its performance.
Transfer Learning
AI agents can leverage knowledge gained from one task to improve performance on another related task. For example, an AI agent trained on recognizing objects in images can use that knowledge to better understand scenes in videos.
Data Augmentation
By generating additional training data through techniques like data augmentation, AI agents can improve their robustness and generalization capabilities. This involves creating variations of existing data to expose the agent to a wider range of scenarios.
These mechanisms enable AI agents to become more accurate, efficient, and capable over time, making them valuable tools in various applications. If you have any more questions or need further details, feel free to ask!
There remain challenges in implementing AI agents in organizations but those are fast vanishing. Poor data or lots of data is one. Most importantly, navigating ethical and regulatory concerns is a big challenge. Cost used to be another hurdle but companies like Microsoft with Co-pilot, etc have overcome this one.
All in all, even though many organizations today do lack the necessary expertise to develop and manage AI agents, it is getting easier.
Reference:
(1) 6 Key Challenges in AI Implementation: Solutions - DataQueue. https://blog.dataqueue.ai/ai-business-where-to-start/ai-challenges-solutions.
(2) The promise of gen AI agents in the enterprise | McKinsey. https://www.mckinsey.com/industries/technology-media-and-telecommunications/our-insights/the-promise-and-the-reality-of-gen-ai-agents-in-the-enterprise.
(3) AI Implementation Challenges: Navigating the Road to Adoption. https://rtslabs.com/ai-implementation-challenges.
(4) How Intelligent Agents in AI Can Work Alone | Gartner. https://www.gartner.com/en/articles/intelligent-agent-in-ai.
(5) 11 Challenges Of Adopting AI In Business (And How To Address ... - Forbes. https://www.forbes.com/councils/forbesbusinesscouncil/2023/10/24/11-challenges-of-adopting-ai-in-business-and-how-to-address-them-head-on/
(6) https://blog.dataqueue.ai/ai-business-where-to-start/ai-challenges-solutions
(7) 185 real-world gen AI use cases from the world's leading organizations. https://cloud.google.com/transform/101-real-world-generative-ai-use-cases-from-industry-leaders.
(8) Why agents are the next frontier of generative AI - McKinsey & Company. https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/why-agents-are-the-next-frontier-of-generative-ai.
(9) 5 Companies Using AI for Customer Service - HubSpot Blog. https://blog.hubspot.com/service/companies-using-ai-for-customer-service.
(10) Microsoft launches AI ‘agents’ to woo investors ahead of earnings. https://www.msn.com/en-gb/money/other/microsoft-launches-ai-agents-to-woo-investors-ahead-of-earnings/ar-AA1sEi5I.
11) Marc Benioff says Microsoft rebranding Copilot as AI 'agents' shows they're in 'panic mode'. https://www.msn.com/en-us/news/technology/marc-benioff-says-microsoft-rebranding-copilot-as-ai-agents-shows-they-re-in-panic-mode/ar-AA1sFSTd.
(12) What Are AI Agents? | Oracle United Kingdom. https://www.oracle.com/uk/artificial-intelligence/ai-agents/.
(13) What Are AI Agents? - IBM. https://www.ibm.com/think/topics/ai-agents.
Disclaimer: Just a heads-up. Remember, "Living With AI" articles are written for the curious everyday folks, not the AI expert. While we try our best to keep things accurate, sometimes, we might (over) simplify things a bit, or leave out some super technical stuff. Think of it like explaining rocket science with a baking soda volcano - fun and fizzy, but not quite the real deal! Don't worry, if you're hungry for more technical details, there's a whole universe of resources out there waiting to be explored.