This article is an excerpt from AI Mastery for Finance Professional.sa new book by Glenn Hopper, published by Readers Press. Copyright © 2024. Mr. Hopper will be the keynote speaker at the Finance & Accounting Technology Expo in New York on October 30th.
OpenAI, a leading research organization and technology company specializing in the development of artificial intelligence, has outlined five steps for artificial intelligence that are very representative of the direction the industry is heading.
Founded in December 2015, OpenAI is focused on how artificial general intelligence benefits all humanity. Known for breakthrough technologies such as ChatGPT and DALL-E, OpenAI aims to advance AI while prioritizing safety and ethical considerations.
OpenAI was originally nonprofit, but moved to a limited-profit model in 2019 to attract larger investments, including about $13 billion from Microsoft as of 2024. The company is focused on the responsible use of technology and actively participates in discussions about the social impact of AI. and address related challenges.
OpenAI's five-step framework for mapping the progress of AI systems toward achieving artificial general intelligence is designed to provide a structured path and milestones as AI technology advances. (See table below.)
OpenAI and other major underlying models are currently focused on moving from Level 1 to Level 2, which is expected to make significant progress as the inference capabilities of AI models are enhanced. Successful development of AI systems along the path outlined by OpenAI's framework could have a profound impact on society.
The Rise of Agentic AI: How Autonomous AI Systems Will Transform Human-AI Interactions
The next frontier in artificial intelligence is the development of agent AI, autonomous systems that can set and pursue goals without direct, ongoing human intervention. The transition from narrow, task-specific AI to more flexible agent systems means that computer systems can move from being tools that require continuous input to performing tasks for us in the digital and (through robotics) physical worlds. Move to a machine that can run it.
Agentic AI refers to artificial intelligence systems that operate, make decisions, and take actions with some degree of autonomy without continuous human intervention. These AI systems are designed to achieve specific goals and can adapt to dynamic environments by learning from experience. Unlike traditional AI, which has limited ability to reason and plan, agent AI behaves like an independent “agent” that can respond to real-time changes in the environment.
A key feature of agent AI is its goal-directed behavior. Self-driving cars are a type of agent AI that navigate roads, make decisions, and adapt to changing traffic conditions without input from a human driver. In the financial sector, agent AI could potentially manage investment portfolios by autonomously making buy and sell decisions based on market conditions and set strategies.
current limit
Today's AI systems, while superior in linguistic capabilities and specialized areas, are still fundamentally passive tools. Chatbots, such as OpenAI's ChatGPT and Anthropic's Claude, can have human-like conversations and assist with a variety of tasks, but they operate on a request-response basis and provide output based on user prompts. .
These systems lack the ability to actively pursue goals, autonomously gather information, and adapt their behavior based on changing circumstances. They are limited by the information contained in their training data and cannot update their knowledge and skills in real time.
| level | explanation | Impact and concerns |
| level 1 Conversational AI/Chatbot |
This stage represents the current state of AI with language models such as ChatGPT and Claude that can have human-like conversations and assist with a variety of tasks. | Concerns remain that inaccuracies, inconsistencies or “hallucinations” in responses could lead to misunderstandings and misinformation. |
| level 2 Human level problem solver/reasoner |
Achieving human-level problem solving across a variety of domains is an important milestone. This includes reducing AI hallucinations, improving accuracy, and ensuring that AI systems can handle complex problem-solving tasks. | At this level, it has the potential to transform industries by automating complex decision-making processes. Concerns include the difficulty of matching human expertise in specific fields and the need to integrate AI with other technologies such as knowledge representation. |
| level 3 agent |
AI systems that can autonomously perform tasks and make decisions over long periods of time are a major step toward artificial general intelligence. | While this capability has the potential to greatly improve the automation of a vast range of tasks, it is important to ensure that these autonomous agents operate safely and reliably in line with human values and avoid unintended consequences. This poses significant challenges to assurance. |
| level 4 innovator |
AI systems that can generate original ideas and push the boundaries of current knowledge have the potential to bring about significant advances in a variety of fields. | While the potential for breakthrough innovation is high, it raises questions about the nature of creativity and whether ideas generated by AI can truly be considered as original as those generated by humans. |
| level 5 organization |
AI systems that manage entire organizations will revolutionize the way companies and institutions operate. | This level can lead to unprecedented efficiency and capability in management. It also raises concerns about the concentration of power in AI systems, the potential impact on human employment, and the challenges in maintaining human autonomy. |
The promise of agenttic AI
In contrast, agentic AI systems will be able to operate with greater autonomy and flexibility. They will be able to set and pursue long-term goals, seek out relevant information, and make decisions based on changing circumstances. Rather than simply responding to user requests, agent AI can proactively provide suggestions, anticipate needs, and even initiate tasks on its own.
Imagine a more powerful Alexa or Siri device that can manage a user's schedule and proactively reschedule appointments based on changing priorities or real-time traffic conditions. In business, agent AI systems can continuously monitor market trends, customer sentiment, and competitor behavior and adapt strategies in real time.
Agenttic AI could also enable more natural and fluid interactions with users. Instead of today's chatbot-like interactions, agent systems can engage in more context-specific, multi-turn interactions to build and maintain consistent conversations over time. It can also learn and adapt to individual user preferences and communication styles.
Challenges and impacts
As AI systems gain more decision-making autonomy, ensuring that autonomous AI systems are safe, reliable, and consistent with human values becomes paramount. Robust “anti-vulnerability” mechanisms for transparency, accountability and oversight will be needed.
There are also potential economic implications. Agenttic AI has the potential to automate more cognitive tasks and decision-making roles, which could lead to significant productivity gains. But it also raises questions about job losses and the distribution of AI benefits.
