AI Strategies and Business Risk Flashcards
What is the “context vector”?
A fixed-length vector that comes from the last state of the encoder, aiming to represent a compressed form of the entire source sentence. This context vector is then used by the decoder to aid in generating the target sentence.
What is the bottleneck issue in sequence-to-sequence models?
Traditional sequence-to-sequence models use a fixed-length context vector to represent the entire source sentence, which can lead to information loss, especially for longer sentences.
How does the Transformer architecture address the bottleneck issue?
Instead of relying on a single, fixed-length context vector, the Transformer employs a weighted sum approach, integrating the context vector (V) into each of the decoding vectors. This allows the decoder to use the information from all encoder outputs, improving accuracy and mitigating information loss.
How does parallel processing on GPUs enhance the Transformer’s performance?
The Transformer’s architecture allows for parallel processing of the entire sequence, unlike traditional NNs that process sequentially. This parallel processing capability is accelerated by modern GPUs, leading to significantly faster training times.
What is Generative AI?
A subset of artificial intelligence that focuses on generating new data that shares the same characteristics as some existing data.
What are some drawbacks of Generative AI?
Limited token processing capacity.
May lack domain-specific knowledge for individual industry problems, necessitating fine-tuning.
What is Algorithm Economy?
Algorithm Economy refers to the business landscape where algorithms are central to value creation. Companies design, build, and leverage algorithms to differentiate themselves, enhance efficiency, and deliver better customer experiences.
Why is Algorithm Economy significant for businesses?
Global companies are increasingly utilizing AI algorithms for value creation, leading to a growing consensus that businesses not exploring or integrating AI might be at a competitive disadvantage in the long run.
What is horizontal AI?
Horizontal AI refers to artificial intelligence
solutions that are designed to provide generalized
capabilities across multiple industries or domains.
What is vertical AI?
Vertical AI solutions are designed for specific industries or tasks.
How is Amazon using AI to enhance sports experiences? Give an example.
Amazon is leveraging AI in sports through data analysis, predictive modelling, and enhanced fan engagement. For instance, in Formula 1 racing, Amazon Kinesis facilitates real-time data processing, which allows teams and broadcasters to gain insights, predict outcomes, improve strategies, and deliver a better fan experience.
What are the 3 main risks associated with using AI?
Data acquisition challenges, lack of explainability, and potential negative customer reactions.
What are the 2 methods of learning used to resolve Data Acquisition challenges?
Reinforcement Learning: Training AI agents through trial and error, rewarding desired actions, and penalizing undesirable ones.
Transfer Learning: Adapting a model trained on a large dataset in one domain to a similar domain with less data available.
What is the “black box” problem in AI, and how does it relate to Lack of Explainability?
The “black box” problem refers to the opacity of complex AI models, where it’s challenging to understand how inputs are processed to reach outputs. This lack of transparency makes it difficult to explain AI decisions, hindering trust and accountability.
What is Explainable artificial intelligence (XAI)?
A set of processes and methods that allow human users to comprehend and trust the results and output created by machine learning algorithms.
