AI

Question 1

What does LLM means?

Answer 1

Large language models (LLMs) make up a class of foundation models that can process massive amounts of unstructured text and
learn the relationships between words or portions of words, known as tokens. This enables LLMs to generate natural-language text,
performing tasks such as summarization or knowledge extraction. Cohere Command is one type of LLM; LaMDA is the LLM behind
Bard.

Question 2

What does GPT means?

Answer 2

Generative Pretrained Transformers

Question 3

Is GPT a general purpose technology?

Answer 3

YES

Question 4

What is the potential impact of GPT at labor market?

Answer 4

Hihg…
- 19% of jobs will have at least 50% of their tasks exposed when considering both current capabilities and anticipated tools to build upon them
- 80% of US workforce will have at least 10% of their tasks exposed
-ocupations with higher salaries will have higher exposure
roles heavily reliant on science and critical thinking skills will be less exposed while roles with skills like programming or writing will be more exposed
- per industry, information processing industries will be more exposed while manufacturing, agriculture, mining, etc. will be less

Question 5

What does the term Singularity refers to?

Answer 5

to the moment when machines equals humans

Question 6

What are the different types of ML?

Answer 6

1. supervised Learning. Discriminative models that are capable of predicting some aspects of the data based on inputs
2. Unsupervised Learning. Like generative AI, are capable of modeling some properties of the data creating new content (text, image, audio, video)
3. Reinforcement Learning, capable of making decisions based on data

Question 7

What are the main risks of AI?

Answer 7

1. Biases, deep fakes and miss interpretation (data curation at scale is challenging)
2. Hallucination, may block critical use cases beyond creative drafting
   (AI dont know how to say “I dont know” which might lead to invent)
3. Costs, lots of energy and high end HW

Question 8

AI developed by layers…

Answer 8

FOUNDATIONAL MODELS (UPPER). Large and expensive AI models of general purpose (takes long time and $$). Hyper scalers best positioned

SPECIALIZED MODELS (MIDDLE). Developed industry vertical use cases based on the above level (less time and investment to develope)

APPS (BOTTOM). Via API, to provide specific services on top of the above

Question 9

What is a strong use case for generative AI nowadays?

Answer 9

Simulators

Question 10

Human-in-the-loop?

Answer 10

Still necessary for a long time

Question 11

What is a Transformer?

Answer 11

A type of neuronal network that uses self-attention mechanisms to process large sequences of data

Question 12

How to assess your company AI needs…

Answer 12

1. Define the objective (with measurable outcomes)
2. Evaluate current capabilities
3. Identify the data needed
4. Select the right AI system
5. Monitor results and reassess

Question 13

Generative AI

Answer 13

Generative AI is a type of AI that can create new content (text, code, images, video) using patterns it has learned by training on extensive (public) data with machine learning (ML) techniques.

Question 14

Foundation models (FMs)

Answer 14

Foundation models (FMs) are deep learning models trained on vast quantities of unstructured, unlabeled data that can be used for
a wide range of tasks out of the box or adapted to specific tasks through fine-tuning. Examples of these models are GPT-4, PaLM 2,
DALL·E 2, and Stable Diffusion

Question 15

Fine-tuning

Answer 15

Fine-tuning is the process of adapting a pretrained foundation model to perform better in a specific task. This entails a relatively short
period of training on a labeled data set, which is much smaller than the data set the model was initially trained on. This additional training
allows the model to learn and adapt to the nuances, terminology, and specific patterns found in the smaller data set.

Question 16

Prompt engineering

Answer 16

Prompt engineering refers to the process of designing, refining, and optimizing input prompts to guide a generative AI model toward
producing desired (that is, accurate) outputs.

Question 17

Hallucinations

Answer 17

…whereby the generative AI model presents an incorrect response based on the highest probability response; the accidental release of confidential personally identifiable information; inherent bias in the large data sets the models use; and high degrees of uncertainty related to intellectual property (IP).