re: Current A.I. explosion…

quote:

---

That’s not AI That’s advanced machine learning based off of complex indexing patterns

---

Interesting. Can ya elaborate on that?

quote:

---

Interesting. Can ya elaborate on that?

---

The distinction between AI and machine learning can be a bit blurry, but generally speaking, AI refers to the broader concept of creating machines that can perform tasks that would normally require human intelligence, while machine learning is a specific approach to achieving this goal.

Machine learning involves training a machine to learn from data and improve its performance at a specific task over time, without being explicitly programmed for that task. This is typically done using algorithms that can identify patterns and make predictions based on those patterns.

The process of generating text using the approach mentioned in the original statement - feeding in articles and training the machine to adopt a specific writing style - is an example of machine learning. Specifically, it involves using natural language processing (NLP) techniques to analyze the text and identify patterns in the data.

So, while the output may seem like it was generated by an AI system, it is actually the result of a specific type of machine learning. The machine is not making creative decisions or exhibiting human-like intelligence, but rather applying a set of rules and patterns that it has learned from the training data.

In summary, the process of generating text using complex indexing patterns is an example of advanced machine learning, but it does not necessarily involve true AI.

quote:

---

So is a search engine if that’s the standard - so is Google maps.

---

All of this “GPT is just a search engine” talk feels like an attempt to bury our collective heads in the sand. GPT isn’t AGI, but it’s a hell of a lot more than a search engine.

The contention seems to be that GPT doesn’t actually “understand” its inputs or outputs - it’s just pattern-matching. That’s a silly contention for a few reasons:

1. Pattern-matching happens to be a pretty key part of how the human brain works as well.

2. True “understanding” is not a particularly easy thing to measure. Before we know it, we will be having the same debate about whether GPT-12 actually “feels” emotions or is just associating emotions with interactions based on its training. See where this is going?

These are questions of sentience, and the fact that they are being debated at all tells you how far the technology has come. At some point it doesn’t matter whether the machine truly understands, or feels, or whatever - we may never be able to know that any more than we can know whether another person sees the color blue the same way we do.

3. While GPT and other large language models don’t meet the generally-accepted criteria for AGI today, it’s not hard to imagine how LLMs are a huge step in that direction. There are signs that GPT is already capable of determining fairly complex cause-and-effect relationships using the same pattern-matching logic that people criticize. It’s already teaching itself new skills without being specifically trained for them, such as when it learned to identify positive and negative connotations without being told to do so.

All of these things represent a lot of “the hard part” with respect to AGI. People seem to be judging LLMs in a vacuum, in the context of search engines and chatbots, without thinking about what these abilities will mean once expanded and connected to other systems. It’s like looking at an internal combustion engine and saying “yeah, but all it does is turn a shaft.”

quote:

---

I classify AI as something we, ultimately, won’t be able to control at all.

---

That sounds like a pretty unconventional way to define it.

Anyway.. /rant