https://example.org/chapter1/s1/index.html TL;DR Too long to read? Prefer to listen to this section? We got you covered! This is a version of this section as an audio podcast produced using Google’s NotebookLM. Using AI to teach you AI, how meta! Your browser does not support the audio element. Alternatively, if you feel like you know this already, try your hand at the optional quiz below and see how you do. Or you can just skip to the next section. We won’t judge you! Hugo en-us https://example.org/chapter1/s1/activity/index.html Mon, 01 Jan 0001 00:00:00 +0000 https://example.org/chapter1/s1/activity/index.html Test Your Knowledge: Introduction to AI and LLMs Let’s see how much you’ve learned! This quiz tests your understanding of the evolution of AI, the 2025/2026 landscape, and the core capabilities and limitations of Large Language Models. --- shuffle_answers: true shuffle_questions: false --- ## What was the key breakthrough introduced by the Transformer architecture in 2017? > Hint: Think about how earlier models processed text compared to Transformers. - [ ] It dramatically reduced the amount of training data needed > Not quite. Transformers actually require vast amounts of training data. Their breakthrough was in how they process that data. - [x] It enabled models to process entire sequences simultaneously and understand context through attention mechanisms > Correct! Transformers introduced attention mechanisms that allow the model to analyze relationships between all words in a sequence at once, rather than processing text sequentially like RNNs and LSTMs. - [ ] It eliminated the need for GPUs in model training > Incorrect. Transformers are computationally intensive and rely heavily on GPU acceleration. - [ ] It allowed models to think and reason like humans > Not accurate. While Transformers produce impressive results, they predict patterns rather than truly "thinking." Reasoning capabilities came later with specialized reasoning models. ## Which of the following best describes the distinction between reasoning models and multimodal models in the 2025/2026 landscape? > Hint: Consider what each model type is optimized for. - [ ] Reasoning models are always larger than multimodal models > Model size doesn't define the category. Some multimodal models have more parameters than reasoning models. - [ ] Multimodal models can only process images while reasoning models only process text > This is an oversimplification. Multimodal models handle multiple types of input/output, and reasoning models focus on deliberative problem-solving. - [x] Multimodal models process multiple types of input (text, images, audio), while reasoning models employ internal step-by-step thinking for complex problems > Correct! Multimodal models like GPT-4o handle different data types natively, while reasoning models like o3 and DeepSeek R1 use internal deliberation to break down complex problems before answering. - [ ] Reasoning models are the next version of multimodal models > Incorrect. These are distinct model categories that serve different purposes, not sequential versions. ## A company's AI chatbot confidently cites "Dr. Sarah Chen's 2024 study on neural scaling laws in Nature" -- but no such paper exists. What type of error is this? > Hint: Consider whether this error stems from training data or from the model's generation process. - [ ] A bias error from the training data > Not quite. Bias errors reflect systematic patterns in training data, not fabricated citations. - [ ] An outdated training data error > Incorrect. The issue isn't that the information is outdated -- the citation was never real. - [x] A hallucination -- the model fabricated a convincing but nonexistent source > Correct! This is a classic hallucination: the model generated a specific, plausible-sounding citation that has no basis in reality. Hallucinations are particularly dangerous because they can be very detailed and convincing. - [ ] A prompt injection attack > No. Prompt injection involves malicious user input manipulating the model's behavior, not the model spontaneously fabricating information. ## A healthcare company is deploying a customer-facing AI chatbot to answer patient questions about symptoms and medications. Which AI limitation should concern them MOST when prioritizing their risk mitigation strategy? > Hint: Consider which limitation could cause the most severe real-world harm in this specific scenario, not just which limitation is most common. - [ ] Bias in training data leading to unfair outputs across demographics > Bias is a real concern for healthcare AI, and biased medical advice could disproportionately affect certain populations. However, in a customer-facing symptom and medication chatbot, the more immediate and severe risk is the model confidently providing fabricated medical information that patients act on -- a single hallucinated drug interaction could cause direct physical harm. Bias mitigation is important but is a systemic issue addressed through training data curation, not the most urgent deployment risk. - [x] Hallucinations -- the model fabricating convincing but false medical information that patients may act on > Correct! Hallucinations pose the greatest risk in this scenario because fabricated medical information (e.g., invented drug interactions, nonexistent dosage recommendations, or false symptom-condition associations) could directly cause patient harm. The evaluation criteria: (1) consequence severity -- acting on false medical advice can cause physical injury or death; (2) difficulty of detection -- hallucinated medical facts often sound authoritative and plausible to non-experts; (3) user trust -- patients tend to trust chatbot responses as vetted medical information; (4) frequency -- LLMs regularly hallucinate specific details like dosages and interactions. While bias and prompt injection are real concerns, hallucination risk demands the highest priority because the harm is direct, immediate, and potentially irreversible. - [ ] Prompt injection attacks that could manipulate the chatbot's medical responses > Prompt injection is a genuine security concern, and a manipulated medical chatbot could provide dangerous advice. However, prompt injection requires a malicious actor to actively exploit the system, whereas hallucinations occur spontaneously during normal operation -- every patient interaction carries hallucination risk without any attacker involvement. For a customer-facing medical chatbot, the baseline risk of hallucinations affecting all users outweighs the targeted risk of prompt injection attacks. - [ ] Adversarial inputs causing the model to behave unpredictably > Adversarial inputs are a valid concern but represent a targeted attack vector requiring sophisticated expertise. In a patient-facing chatbot, the far more pressing risk is hallucinations that occur naturally during routine use. Every conversation carries the risk of fabricated medical information, while adversarial attacks require deliberate, skilled intervention and affect fewer interactions. ## Small Language Models (SLMs) like Phi-4 and Gemma 3 are significant in the 2025/2026 landscape because they: > Hint: Think about what problem SLMs solve that large models cannot. - [ ] Are always more accurate than large models > Incorrect. SLMs trade some capability for efficiency -- they excel at focused tasks but may not match frontier models on the most complex problems. - [ ] Require no training data to function > All language models require training data. SLMs are trained on carefully curated datasets. - [x] Enable powerful AI capabilities on devices without internet connectivity, at a fraction of the cost and latency > Correct! SLMs like Phi-4, Gemma 3, and Llama 3.2 (1B-3B) can run on phones, laptops, and edge devices, bringing AI to privacy-sensitive and low-connectivity scenarios. They often match much larger models on focused tasks. - [ ] Have completely replaced large language models for all use cases > SLMs complement rather than replace large models. They excel at specific tasks, while frontier models handle the most complex analysis. ## Which of the following is NOT a vulnerability or limitation of current LLMs? > Hint: Consider which item doesn't represent a genuine risk or weakness. - [ ] Biases inherited from training data that can lead to unfair outputs > This is a real limitation -- LLMs can reflect and amplify biases present in their training data. - [ ] Susceptibility to prompt injection attacks where malicious inputs manipulate outputs > This is a real vulnerability -- prompt injection is a well-documented attack vector against LLMs. - [x] The ability to autonomously update their own weights during conversations to improve over time > Correct! This is NOT a real capability or limitation. LLMs do not update their weights during inference (conversation). Weights are fixed after training. This is a common misconception. What appears as "learning" in conversation is actually context management within the conversation window. - [ ] Generation of hallucinations -- fabricating convincing but false information > This is a real limitation -- hallucinations remain one of the most significant challenges with LLMs. ## The evolution of AI can be summarized as progressing through distinct stages. What is the correct ordering? > Hint: Think about how each stage built on the previous one's capabilities. - [ ] Deep learning, then rule-based systems, then machine learning, then Transformers > The order is wrong. Rule-based systems came first, predating machine learning. - [ ] Machine learning, then Transformers, then deep learning, then rule-based systems > Incorrect. This reverses the actual progression. - [x] Rule-based systems, then machine learning, then deep learning, then Transformers and generative AI > Correct! AI evolved from rigid rule-based systems to data-driven machine learning, then to deep learning with neural networks, and finally to the Transformer architecture that enabled the generative AI revolution we see today. - [ ] Transformers, then deep learning, then rule-based systems, then machine learning > This reverses the actual historical progression entirely.