AI.5 Create Topic Sentences For Each Topic Card In The Essay
Start your paragraph with a sentence which states the paragraph’s topic.
The next step in developing my scaffold was to write a topic sentence for each paragraph topic (Section 2.11). When creating topic sentences, I perform the first ‘proper writing’ in my manuscript, because so far I have avoided complete sentences. However, I also realize my first sentences are first drafts. I try not to waste time writing perfect sentences. I try to generate topic sentences as quickly as possible because I will improve my wording later.
I wrote topic sentences for my essay as follows: I took each topic card in order. I read the topic on the blank side of the card. I turned the card over, and on the top of its (lined) side I wrote a topic sentence. I did so while keeping the card’s topic in mind; my goal was to write a complete sentence which expressed the topic. Thus, the topic on one side of the index card scaffolded creating the topic sentence on the card’s other side.
As noted earlier, a paragraph’s topic sentence supports a manuscript’s narrative structure. After creating topic sentences, I felt I could obtain a stronger sense of my paper’s narrative by reading the topic sentences in order. When I read my first set of topic sentences for my essay, I found my narrative disappeared toward the end. As described in Section AI.4, my problem narrative made me return to working with topic cards; I removed one topic card and added four new topic cards in the same location, before I was satisfied with my narrative.
Table AI-3 provides, in order, each topic sentence for each paragraph in my essay – at least for the version of the essay represented in my topic cards:
| Table AI-3. The topic sentence generated for each topic card (each paragraph) in the essay. | |
|---|---|
| Paragraph | Topic Sentence |
| 1 | We live in an artificial intelligence (AI) revolution fueled by a new invention called a large language model (LLM). |
| 2 | LLMs are revolutionary because they can generate long, detailed, meaningful responses to short text prompts. |
| 3 | LLMs’ performance has generated many questions in both the popular press and scholarly journals. |
| 4 | Speaking as a cognitive scientist, I feel such questions miss the key point. I am interested in the question ‘Can LLMs inform cognitive science?’. |
| 5 | Below, I argue LLMs may indeed be able to inform cognitive science – but only if researchers expend considerable effort to study the internal structure of LLMs in order to discover how LLMs produce their amazing behavior. |
| 6 | Modern AI’s excitement and controversy comes from an LLM’s ability to generate paragraphs of meaningful sentences in response to short prompts or questions. |
| 7 | Cognitive science has studied human language for decades. Cognitive science’s most influential account proposes human language involves specialized rules or processes manipulating complex mental representations of sentences. |
| 8 | Cognitive scientists who believe human language is the rule-governed manipulation of symbols do not believe LLMs inform cognitive science. |
| 9 | UC Berkeley psychologist Steven Piantadosi agrees LLMs do not use grammatical rules. However, he then proceeds to argue an LLM’s high level performance without using rules refutes Chomskyan linguistics. |
| 10 | My own research concerns cognitive science’s foundations, with particular interest in the relation between theories based on rules and symbols and theories based on artificial neural networks. I therefore recognize as historical precedent for Piantadosi’s position. |
| 11 | In the mid-1980s, cognitive science found itself in the midst of what is now called its connectionist revolution. |
| 12 | The rise of multilayer perceptrons in cognitive science caused a revolution because proponents of artificial neural networks attacked traditional theories which appealed to the rule-governed manipulation of symbols. |
| 13 | My interest in the connectionist revolution focused on a curious aspect of the revolutionaries’ argument: they assumed networks abandoned symbols and rules, but never provided evidence to support their assumption, or to show what their networks used to replace symbols and rules. |
| 14 | Surprisingly, when I looked inside my trained networks, I discovered lots of structures which resembled theories based on symbols and rules. |
| 15 | Importantly, I did not usually find network structure which replicated existing formal theories. Instead, I found new formal structures which could inform a cognitive science based on symbols and rules. |
| 16 | I feel my research on interpreting the structure of artificial neural networks demonstrates the peril of gee whiz connectionism. |
| 17 | The moral of my story about my own work is my suspicion LLMs will only inform cognitive science when researchers abandon mere assumptions about what makes LLMs different from rule and symbol models, and instead see evidence about both the similarities and differences between both types of models. |
| 18 | Why must we look inside LLMs to inform cognitive science? Cognitive scientists have long known psychologically plausible performance can be produced by methods completely unrelated to the processes of human cognition. |
| 19 | Indeed, I strongly suspect LLMs use methods radically different from those used by humans because they represent stimuli and responses with encodings unrelated to any proposed by cognitive scientists. |
| 20 | Because I suspect LLMs use methods unrelated to human cognition, I also believe Piantadosi’s (2023) claim ‘LLMs refute Chomskyan linguistics’ illustrates the danger of being seduced by network performance. Again, cognitive scientists recognize human-like performance is not sufficient to establish human-like processing. |
| 21 | LLM proponents recognize LLMs use some method to produce a remarkable facility with language. |
| 22 | However, understanding how LLMs convert stimuli into responses is extremely challenging, because LLMs are intimidatingly large and complex systems. |
| 23 | Fortunately, researchers recognize the need to extract potentially novel theories or representations from LLMs and are developing new techniques to understand a LLM’s internal structure. |
| 24 | My hope is more work of this sort is on the horizon. |
| 25 | Piantadosi (2023, p. 30) claims “large language models rewrite the philosophy of approaches to language. Do LLMs refute Chomsky’s approach? Do LLMs represent a new connectionist revolution for cognitive science? I believe we can’t answer such questions – yet. |
