AGI Defined

Artificial General Intelligence (AGI)

  1. The ability of a machine (e.g., a computer) to exhibit "intelligence" in a "general" manner
    1. Depends on what "intelligence" means.
    2. Depends on what "general" means
    3. Thus, since an LLM or related technology can "seem smart", e.g. "write an essay", "summarize text", "produce an argument for a given position", "create a drawing", or "make a podcast audio episode recording file based on the material in some reference documents", or "create a software program to meet some given requirements" and if said technology system can do this on roughly "any topic" (such as: literature, economics, politics, fiction, science, mathematics, computer science, etc.), then some people will argue that this constitutes "artificial" (man-made computerized) "general" (across fields, domains, media, and activity types) "intelligence" (assessing and putting information together in new ways, including weighed decisions on what paths to take).
    4. Such an AGI can still have all the biases and limitations intrinsic to its programming and training and context.
  2. A specific capability whereby a machine is able to reason in a mechanical manner by applying brute "intelligence" (ability to reason) to a situation, with the end result that the machine is "as smart as a machine would be imagined to be", that is to say, able to broadly apply all available inputs rationally, without succumbing to the typical pitfalls that humans tend to fall into, since it is a machine ("artificial") and able to reason broadly ("general") and capable of applying the information that it is able to observe ("intelligence")
    1. Such an AGI is anticipated to be capable of being relatively "bias free", being artificial
    2. Such an AGI is anticipated to be capable of reasoning abstractly; that is, not depending upon having seen an exact pattern before, and able to take on "new learning" to become more "capable" over time, though at some point it may reach an asymptote of more or less "true objectivity"
    3. Such an AGI is anticipated to be able to ingest basically arbitrarily large datasets (e.g., the internet or a library) and create intellectual outputs that reflect essentially the "objective knowable truth" determinable from said datasets
      1. Specifically, the "truth discernment" capability, which is a hallmark of intelligence and reasoning, stands out in this definition
      2. Specifically, a "non AGI system" that may otherwise appear close to AGI, might produce similar intellectual outputs based on similar available datasets and interactions, but would rely on faulty shortcuts such as "quantity" of arguments for a position (availability bias, frequency bias) or "reputation" of proponents of a position (appeal to authority, star power), or textual "closeness" (proximity) of information, or other such factors, in order to evaluate levels of truth (in domains with various conflicting positions), instead of evaluating those points on their "actual" symbolic meaning and logical consequences.
        1. This limitation is a key factor in differentiating "what people mean" when they claim that "AGI has not been achieved" -- no system meeting this capability is "known" to have been made publicly available as of February, 2026.
  3. A machine such that "Anything we can do, they can do better. They can do anything better than us." (ref)
    1. "Let an ultraintelligent machine be defined as a machine that can far surpass all the intellectual activities of any man however clever. Since the design of machines is one of these intellectual activities, an ultraintelligent machine could design even better machines; there would then unquestionably be an 'intelligence explosion,' and the intelligence of man would be left far behind. Thus the first ultraintelligent machine is the last invention that man need ever make." - Irving John Good, 1965
  4. If it can make money by itself.  (Give it funds, then it develops a business that grows those funds by itself.)
  5. "Agentic is AGI."
    1. "If a woman gets married at 25 and her kid is 25, how old is she?...This is what LLMs are dealing with. You dont tell them everything they need to know and they are left to fill in the gaps. Which may, and sometimes often means they lie....That's what Agentic does differently, it'll go find the gaps before answering." (ref)
  6. "artificial general intelligence (AGI)—by which we mean highly autonomous systems that outperform humans at most economically valuable work" (ref. OpenAI's charter)