"The Near Future of AI: Action-Driven Innovation and the Power of Sharing Radical Ideas"


Hatched by Glasp

Aug 14, 2023

4 min read


"The Near Future of AI: Action-Driven Innovation and the Power of Sharing Radical Ideas"

In today's rapidly evolving technological landscape, artificial intelligence (AI) continues to push boundaries and redefine what is possible. One exciting development in the field is the emergence of action-driven AI models like the ReAct model (Yao et al. 2022, arxiv). This model takes a three-step approach: Thought, Act, and Observation, allowing it to make choices and learn from the outcomes of its actions. By incorporating cognitive assets such as search, these action-driven models have the potential to revolutionize various industries and pave the way for what some may consider to be true artificial general intelligence (AGI).

LLMs (large language models) have already demonstrated their prowess in question-answering tasks, particularly when prompted to "think step by step" (Kojima et al. 2022, arxiv). However, their performance can be further enhanced by leveraging external cognitive assets. By accessing data from external sources, LLMs can bridge the resource gap and provide even more accurate and comprehensive answers. This approach aligns with OpenAI's 002-text-davinci model, which combines instruction tuning and Reinforcement Learning from Human Feedback (RLHF) to achieve impressive results (blogpost). The potential for reinforcement learning to train AI systems to improve their performance based on specific metrics of interest holds great promise for future advancements.

While established players like OpenAI are at the forefront of AI innovation, there is also room for startups to make a significant impact. Startups that can create powerful feedback loops by solving customer pain points, collecting data, training their models, and iterating are poised for success. This iterative process, akin to building a moat around their AI capabilities, allows them to continuously improve and refine their offerings. As AI agents become more domain-general, the possibilities for automation and new offerings expand, propelling the field forward and opening doors to new opportunities.

In the realm of radical ideas and innovation, the story of Marie Curie serves as a powerful inspiration. Curie, a pioneering physicist and chemist, was known for her groundbreaking work on radioactivity. Three key principles can be gleaned from her life and achievements: embracing the unknown, making radical bets, and sharing ideas.

Curie's willingness to embrace the unknown was evident in her early career when she put forth her daring hypothesis that caused a scientific stir. Instead of shying away from the unfamiliar, she leaned into it, leading to remarkable discoveries and advancements. This mindset of embracing the unknown is a valuable lesson for innovators and AI researchers alike. It encourages us to push boundaries, challenge conventional wisdom, and explore uncharted territories.

Making radical bets is another characteristic that propelled Curie's career. Sometimes, the most significant bet we can make is the bet we make on ourselves. Having confidence in our abilities and taking calculated risks can lead to transformative breakthroughs. Whether it's pursuing a new research direction or venturing into uncharted business territories, embracing the spirit of Curie's radical bets can lead to remarkable outcomes.

Furthermore, Curie's willingness to share her ideas was instrumental in her success. Instead of hoarding knowledge, she actively collaborated with others, including people in the healthcare industry. This openness allowed for a free exchange of ideas and cross-pollination of insights, ultimately driving progress in her field. In the world of AI, sharing ideas is equally important. By engaging with a diverse community of researchers and practitioners, we can gain fresh perspectives, challenge our assumptions, and foster innovation.

To apply these principles to our own lives, we must be willing to face an uncertain future, have confidence in ourselves, and surround ourselves with a community that encourages open exchange of ideas. Taking small risks to improve our work or personal lives, aligning our decisions with our values, and sharing our ideas, even when they are not fully developed, can lead to significant growth and impact.

In conclusion, the near future of AI is action-driven, where models act as agents, making choices and learning from the outcomes. Leveraging external cognitive assets and reinforcement learning will be crucial in driving advancements in AI capabilities. Startups that can create powerful feedback loops and continuously iterate will have a competitive edge. Furthermore, embracing the unknown, making radical bets, and sharing ideas are key principles that can fuel innovation and drive progress, as exemplified by the remarkable life and achievements of Marie Curie. By incorporating these principles into our own lives and work, we can pave the way for transformative breakthroughs and shape the future of AI and beyond.

Actionable Advice:

  • 1. Embrace the spirit of the unknown by exploring uncharted territories and challenging conventional wisdom. Push boundaries and be open to unexpected discoveries.
  • 2. Make radical bets on yourself and your ideas. Have confidence in your abilities and take calculated risks that can lead to transformative breakthroughs.
  • 3. Foster a culture of sharing and collaboration. Engage with a diverse community of researchers and practitioners, actively seek out opportunities for knowledge exchange, and be open to feedback and new perspectives.


  • Yao et al., "The Near Future of AI is Action-Driven", 2022, arxiv.
  • Kojima et al., "LLMs and the Power of Thinking Step by Step", 2022, arxiv.
  • OpenAI Blogpost on Reinforcement Learning from Human Feedback (RLHF).

Hatch New Ideas with Glasp AI 🐣

Glasp AI allows you to hatch new ideas based on your curated content. Let's curate and create with Glasp AI :)