This article is part of the TechXchange: Generating AI.
It seems that artificial intelligence and machine learning (AI/ML) is becoming as ubiquitous as the internet. Even the U.S. Congress is talking about it, or at least around it. Bad actors use AI/ML to create more nefarious attacks while companies are incorporating it into their products and algorithms. People utilize it daily and often unknowingly to crop someone out of a photo or when searching for someplace to eat.
Much of what AI/ML does these days is improve on tasks that are performed daily, but faster and often better. The photo of my aquarium has two added fish (see figure). One was from another photograph while the other was created using generative AI. Do you know which is which? I used the open-source Gimp image editor to create the final photo so that the two Siamese Fighting fish stand out. However, I could have done more to make them blend in or even used AI tools to accomplish that.
This type of graphic manipulation is only the tip of the iceberg, but one that almost anyone can use. The other tools garnering major interest are chatbots like OpenAI’s ChatGPT and Google’s Bard. A number of issues have arisen from the source of training material for these large-language-model (LLM) tools regarding bias and explainability. There are also discussions about legal, moral, and liability issues when it comes to using any output associated with these tools.
These days, we at Electronic Design are making sure we annotate the source of our images. However, that’s not the case for many websites. Likewise, we’re not using generative AI to create articles as some sites are doing. Some of those sites indicate when AI is utilized, though not always. Do you care?
Engineers may be able to overlook these issues if it were limited to images, videos, and text, but the basic AI/ML technologies are part of the software we use to create applications and products from chips to software. Eventually, these aspects of the tool will not be something that can be avoided as these models and technologies are intimately included within the tool.
The technology in general has the potential to vastly improve the output and capabilities of engineers and programmers. Nevertheless, like the rise of security tools and technologies, taking advantage of AI/ML will require a better understanding of the capabilities and liabilities of AI/ML-based tools. Stay tuned for more on this subject.
Read more articles in the TechXchange: Generating AI.