In recent years, the most common use for AI is to make our daily lives easier. When used correctly, it can give us results that are as close as possible to what we need. However, it is important to remember that the use of AI in any setting is a two-way street – its output is only as good as the prompts or input it gets.
In the third episode of the CrushBank AI for MSPs podcast, CrushBank co-founder and Chief Technology Officer David Tan walks us through the good, the bad, and the sometimes somewhere-in-between output that AI can deliver, as well as what can be done to truly harness its potential.
Anyone can be an artist – with a little help
If your perceived lack of artistic talent has kept you away from using Adobe Photoshop in the past, then now is the best time to give it another go.
In the past, only masters of artsy manipulation could get away with clean photo editing work. Tan demonstrates how AI has changed the game for many with Adobe Photoshop’s Generative Fill feature. He simply selected the image he wants to remove, which, in this case, is the tree on the upper-left side of the photo.
After drawing around the tree, Tan selected the tree and typed in “remove tree.” After a few seconds, the tree has been expertly removed from the image with minimal effort.
|Tan tries this again using the same steps, this time to remove the flowers at the bottom of the photo and place a lake instead, once again resulting in a perfectly cleaned up picture.|
|As a finishing touch, Tan also changed the main subject in the photo – the house in the center – and swapped it for a log cabin, creating a more cohesive vibe for the scenery – all done with the help of AI.|
The limitations of AI
Many tasks may have been simplified by AI. However, it still has its limitations. While it takes in data to continuously learn and improve, it is still only as good as the commands it receives. To prove this point, Tan gamely uses his own photos as an example.
|To further enhance his appearance in the photograph, he entered a command to add more hair and ended up with hilarious results.|
|In this case, he had no point of reference as to what the AI was trained on, which explains the curveball that he was thrown.|
A world of possibilities
Tan shared that this experiment helps us visualize what we can potentially do with the help of AI. For example, think of employees who work in a creative field where the use of both visual and written output is crucial. It is rare to find people who can do both types of output well. But with the help of AI, members of a creative team that excel in writing can use their skills to use AI and create visual output that they would usually need to have someone else execute.
Proper AI governance is the key to success
While AI is a gamechanger in many industries, it would not be half as successful if not for the human intervention necessary to guide its output. Working with AI isn’t just about typing up prompts and praying for the best – a lot of work goes into creating a properly working model. And a governance process is needed to avoid the pitfalls. AI governance refers to the policies and practices put in place to ensure responsible and ethical development and use of AI and the underlying LLMs. Its goal is to allow the use of AI, while mitigating risks, protecting individual rights, and creating accountability. AI governance is crucial as AI systems become increasingly integrated into all aspects of business and personal life.
Managed Services Providers (MSPs) use a variety of software applications in their technology stack. Many of these are embedding generative AI into their systems, underpinned by Large Language Models (LLMs). In addition, their employees are making use of tools such as OpenAI and ChatGPT in their day-to-day jobs. Generative AI can do amazing things, but the technology has spread so quickly, often without MSPs being clear on what they are actually using, that they may be exposed in several ways. The MSPs need to make sure all output is trustworthy for practical use. Whether you’re in managed services or a managed services provider, the responsibility to help your clients depends on you.
The good news is that CrushBank uses IBM WatsonX AI, which is built to enable and facilitate governance and control. WatsonX is different from other AI platforms as it allows transparency and control of the LLM environment and uses. The watsonx.ai component is used for building foundation models, generative AI and machine learning. watsonx.ai is used to train, validate, tune and deploy foundation and machine learning models. It can be used with different LLMs. watsonx.governance creates responsibility, transparency and explainability in the data and AI workﬂows. This solution helps you direct, manage and monitor your AI activities.
At the end of the day, it’s all about utilizing AI responsibly.
Listen to the full podcast here.