# AGI, Simply Explained #64 Aug 18, 2025 LLMs https://www.aibutsimple.com/p/agi-simply-explained When people normally talk about Artificial Intelligence (AI), they typically mean AI tools that are becoming increasingly commonplace, such as LLM chatbots like ChatGPT or AI agents such as GitHub Copilot. These tools are all an effort that leads us to Artificial General Intelligence (AGI), which refers to AI systems that can understand, learn, and perform tasks that we would normally expect only a human would be able to do. Although there are indeed many AI agents available nowadays, they are typically built to be good at one specific task or function. ![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,width=1200,format=auto,onerror=redirect,quality=80/uploads/asset/file/1d3efd00-867c-42c5-a8e6-b16ae4107368/Untitled_design__1_.png) AGI, on the other hand, would theoretically be flexible enough to switch between many different tasks while also being smart enough to adapt learned patterns from one task to solve another. To paint a broader picture, imagine if the same AI that could beat a grandmaster in chess could then also help optimize your calendar, construct recipes based on the ingredients in your pantry, and even provide therapeutic comfort when you type your feelings out to it as a chat. ![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,width=1200,format=auto,onerror=redirect,quality=80/uploads/asset/file/161f77f8-0900-441d-855f-9d2bbb1075c3/Untitled_design__2_.png) This broad applicability is why we consider AGI to be considered “general” intelligence. In theory, it would be able to understand, improvise, and learn across a span of topics, much more like humans do. --- ### This guide could help keep your whole organization safe. You wouldn’t allow unmanaged devices on your network, so why allow unmanaged AI into your meetings? Shadow IT is becoming one of the biggest blind spots in cybersecurity. Employees are adopting AI notetakers without oversight, creating ungoverned data trails that can include confidential conversations and sensitive IP. Don't wait until it's too late. This [Shadow IT prevention guide](https://fellow.ai/resources/shadow-it-guide/?utm_campaign=XIAEVJHTGJ&utm_source=beehiiv&_bhiiv=opp_ed8523a6-e522-42c1-963a-4f3dcd18183f_214b5bca) from Fellow.ai gives Security and IT leaders a playbook to prevent shadow AI, reduce data exposure, and enforce safe AI adoption, without slowing down innovation. It includes a checklist, policy templates, and internal comms examples you can use today. [Download the Shadow IT Prevention Guide](https://fellow.ai/resources/shadow-it-guide/?utm_campaign=XIAEVJHTGJ&utm_source=beehiiv&_bhiiv=opp_ed8523a6-e522-42c1-963a-4f3dcd18183f_214b5bca) --- So, is this just science fiction? Or is it our inevitable future? Well, we don’t fully know yet. Although the rise of AI has been exponential in recent years, AGI’s feasibility is still a large question—can AI really replace humans? Today, researchers are exploring several avenues to develop AGI, such as scaling up powerful AI models (like ChatGPT), arguing that consistent development of their neural networks may allow these models to acquire more general human abilities. - A community-made sample timeline of our progression towards AGI can be found [here](https://ai-timeline.org/?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained). There are some nice visuals and explanations! Others are aiming to design neural networks that follow the structure of the human brain, aiming to allow models to make connections the same way humans do, increasing the scope for transferable knowledge. Some believe that AGI will develop to its true potential only when we let AI models interact with the physical world as robots. Well, what’s holding us back from entering the new era of intelligent robots coexisting with humans? Simply put, although AI agents are incredibly useful and have sped up many processes, they simply are not consistently smart. - AI agents can help us solve math equations or write essays, but they are prone to making errors or even lacking logic in discussion. ![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,width=1200,format=auto,onerror=redirect,quality=80/uploads/asset/file/fefa2ac2-b4a2-446c-ad94-c76571078d21/Untitled_design__3_.png) Humans combine reasoning, memory, creativity, and common sense. Although AI has its strengths, it cannot really combine these qualities that make us, well, human—it’s simply not there yet. These models, due to their very nature of learning, are built to be specialized. It remains a challenge to create one that can transfer relevant skills across different domains. For instance, imagine you trained an AI to find and count how many dogs were in a photo. You could ask the same AI to then count how many cats appear in the same photo and notice it struggles to adapt, even though it’s still searching for beings that walk on all fours and have tails. - The transferability of knowledge simply is not there for AI, whereas it is a very human trait to use prior knowledge to make sense of something new. Inevitably, this leads us to [Humanity’s Last Exam](https://agi.safe.ai/?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained), a language model benchmark of over 2,500 questions across a vast array of topics. It was developed to be one of the most challenging benchmarks for AI models, with questions submitted and filtered by subject matter experts. The best AI models at the time of writing still score around 25-30%, whereas human experts hover closer to the 90% mark. ![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,width=1200,format=auto,onerror=redirect,quality=80/uploads/asset/file/fc049161-f16f-4280-8bf2-c3896aa93a8a/unnamed.png) This leads to the last but arguably most critical issue of the AGI conversation, which is that of safety. Yes, we are dealing with a vision that could reshape human life for the better. Imagine a world in which every child has a personal tutor that can teach them every subject, depending on the child’s respective strengths in each class. However, we are still talking about creating something smarter than us that can think for itself and make decisions potentially for its own best interest. Are we confident that we can handle a world where machines are more powerful and effective than humans across the board? We’re potentially looking at large-scale job displacement, if not cases of misuse in the realms of politics and warfare. --- ### [Level Up Your Career With AI, But Simple PRO](https://www.aibutsimple.com/upgrade?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained) ![](https://media.beehiiv.com/cdn-cgi/image/fit=scale-down,width=1200,format=auto,onerror=redirect,quality=80/uploads/asset/file/281697a8-c915-435c-a68b-43374b671698/aibutsimple_academy.png) Supporting us by purchasing a PRO membership will get you: - Exclusive Hands-on Tutorials With Code - Full Math Explanations and Beautiful Visualizations - Ad-Free Reading for an Uninterrupted Learning Experience [Break into AI or sharpen your edge by joining today!](https://www.aibutsimple.com/upgrade?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained) --- As AI keeps scaling, the likelihood that AGI becomes a reality grows, but to some researchers it’s all just a fever dream. In their opinion, human intelligence and its broad-scale applicability to different, concurring problems is a unique gift and something that cannot simply be computed. The research will still carry on due to the potential of AGI, but it is important we continuously study the field and problem systematically in order to ensure that regardless of what we end up building, it remains in the best interests of us all. --- Here’s a special thanks to our [biggest supporters](https://www.aibutsimple.com/upgrade?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained): - Sushant Waidande - Sai Krishna Pashapu --- If you enjoy our content, consider supporting us so we can keep doing what we do. Please share this with a friend! Want to reach 50000+ ML engineers? Let’s work together: [WORK WITH US 📈](https://www.passionfroot.me/aibutsimple?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained) Feedback, inquiries? Send us an email at [edwin@aibutsimple.com](mailto:edwin@aibutsimple.com). If you like, you can also donate to our team to push out better newsletters every week! [Support us ☕](https://ko-fi.com/aibutsimple?utm_source=newsletter.aibutsimple.com&utm_medium=referral&utm_campaign=agi-simply-explained) --- That’s it for this week’s issue of AI, but simple. See you next week! —AI, but simple team