Directions for teachers:

To engage students before reading the article, have them answer the “Before Reading” questions as a warmup in class or for homework. If you’d like to address AI technology in depth, consider combining this assessment with a lesson plan about ChatGPT. Then, ask students to read the online Science News article “How artificial intelligence sharpens blurry thermal vision images” and have them answer the “During Reading” questions. As an optional extension, have students apply what they’ve learned and discuss the “After Reading” questions. A version of the article, “How artificial intelligence sharpens blurry thermal vision images,” appears in the September 23, 2023, print issue of Science News. Science News Explores offers another version of the same article written at a middle school reading level.

Directions for students:

Read the online Science News article “How artificial intelligence sharpens blurry thermal vision images” and answer the following questions as directed by your teacher.

Before Reading
1. If given the opportunity, how likely is it that you’d choose to ride in a self-driving car? Would you feel more safe, less safe or equally safe riding such a car at night vs. at day? Briefly explain your reasoning.

Answers will vary. Some might say they are unlikely to ride such a car, and that riding at night would feel less safe due to limited daylight. However, others might reason that the roads might be less busy at night, therefore they’d feel safer.

2. What is a piece of equipment or technology that a self-driving car would require, but that would be optional or absent in a normal car?

Answers will vary. An example would be sensors, especially cameras or other devices, that monitor the road. Some regular cars do have cameras, but they are not as crucial as they would be in a self-driving car.

During Reading
1. Thermal cameras work by detecting heat sources. To do that, they must sense particular wavelengths of light that our eyes cannot see. What kind of light do thermal cameras detect?

Thermal cameras detect infrared light.

2. Regarding thermal imaging, what is “ghosting?” What problem does ghosting cause?

Ghosting occurs when heat from an object overwhelms image details, such as textures. The problem with ghosting is that it can cause images to appear blurry.

3. When scientists paired artificial intelligence (AI) with thermal-imaging technology, the technique produced more detailed images than produced by the thermal camera alone. What information did the AI reveal that the thermal camera on its own did not?

Artificial intelligence helped untangle details in a thermal image about the textures and types of materials that the viewed objects were made from. 

4. What extra or higher quality data might be available to self-driving cars using the type of AI described in the story?

Self-driving cars might better be able to measure distances and more accurately navigate driving at night.

5. Using current self-driving car technology, how might having many self-driving cars on the same road “confuse one another?” Why do researchers say AI technology could make self-driving car technology safer to scale up?

Currently, self-driving cars judge distance by bouncing signals off of nearby objects. However, problems may occur if many self-driving cars are on the same road because they may confuse another vehicle’s signals for their own. The new technology could reduce this problem because it does not need to emit signals to judge distances.

6. Researchers point out that despite the new camera’s potential, it’s unlikely to appear in vehicles in the near future. Give three reasons why.

Firstly, this new camera is too big for practical use in a vehicle. Secondly, the camera is too expensive. Thirdly, the image processing time needs to be faster in a self-driving car that must respond to situations rapidly.

7. Besides self-driving cars, in what other technology does Fanglin Bao hope to see this AI technology used?

Fanglin Bao hopes to see this technology used in robots.

After Reading
1. Besides image analysis, what is another existing technology or technique that AI might improve? Explain your answer. In your example, does AI serve more to overcome limitations of the technology? Or does it elevate this technology, offering it new capabilities?

Answers will vary. An example is genealogy, both family tree-related and forensic genealogy. In these cases, AI might help dive through loads of information. The machine-learning capabilities might help make sense of mixed data types, such as maps of family trees, DNA analysis sites, public records and more. And this could help identify potential criminals or find family. In the example in the story, AI serves more to address limitations of genealogy techniques by detecting patterns among mixed data from varying sources.

2. A dichotomy is when we think of two things as having a rigid division between them (good vs. evil, for instance.) Dichotomies are sometimes used as literary devices or models to explain contrasted concepts. For example, in the article, Bao suggests a dichotomy that exists between day and night. Dichotomies sometimes oversimplify the contrasted ideas, however, making the concepts seem more distinct than they really are. Besides those already mentioned (day vs. night & good vs. evil), give an example of a dichotomy you’ve seen used either in science or in other aspects of life. How might such a dichotomy, if given only passing thought, oversimplify an otherwise complex pair of concepts?

Answers will vary. Examples might include positive vs. negative, such as in regard to electrical charges. In this example, it’s crucial to understand that there are degrees of positive and degrees of negative. Furthermore, the dynamics between elementary particles derive from degrees of difference between interacting particles. Other possible examples: acid/base, male/female, nice/mean, etc.