Question

a bank is developing an algorithm that will help them decide who to make loans to. as input the algorithm will consider information about the per for the loan and the amount of money requested, and as output the algorithm will decide whether the bank should give that loan. the bank intends to develop the algorithm using machine learning techniques. the algorithm will be trained using data from past loan decisions ma human bankers. which of the following best describes whether this algorithm will include bias? machine learning algorithms cannot be developed using biased data so if there is bias in the data it will be impossible to develop the algorithm the algorithm will not be biased because using machine learning eliminates human biases the algorithm will likely reflect the human biases in the data used to train it while the algorithm may be biased, the eventual decision made by the algorithm will not be

Explanation:

To determine the correct option, we analyze each choice:

1. The first option is incorrect because machine learning algorithms can be trained on biased data, and such data can lead to biased models.
2. The second option is wrong as machine learning models trained on data with human biases (from past human banker decisions) will not eliminate those biases; they can learn and perpetuate them.
3. The third option is correct. Since the algorithm is trained on data from past loan decisions made by human bankers, if those human decisions had biases (e.g., discriminatory lending practices), the algorithm will learn and reflect those biases.
4. The fourth option is incorrect. If the algorithm is biased (due to biased training data), its decisions will also be biased.

Answer:

The algorithm will likely reflect the human biases in the data used to train it