A Newbies Guide to Generative AI

Alright, it can’t just be me,?Artificial Intelligence (AI) or Generative AI has become a buzzword in recent months, right? I have been devoting a lot more time trying to catch up and learn, so in this edition of Conn Talks, I'll provide a newbie's guide to explaining its fundamental principles and discussing the concept of Generative AI, so, let's dive in...

Honestly, I think it had to be around November or December of 2022, I was sitting at the in-laws and my wife’s dad asked me if I had downloaded ChatGPT yet. I honestly had zero clue what he was talking about, but since that time, AI seems to be popping up everywhere, at work, at home and even at conversations over refreshments at the Calgary Stampede last week. ?

Now, since that time I've tried using or reading up on applications like ChatGPT, DALL-E, SnapChat AI Bot, Bing Chat or Microsoft Co-pilot (which Microsoft will be charging $30 per user for E3, E5, Business Standard and Business Premium customers) and i'm sure many more to come. But while using this tech, i'm not really sure I understood how it actually worked. Some of these had free trial period, but most are moving to a pay to use model in some iteration or another.

What exactly is this Generative AI that is promising to transform industries and revolutionize our daily lives?

According to Gartner, who is a trusted source for me:

"Generative AI can learn from existing artifacts to generate new, realistic artifacts (at scale) that reflect the characteristics of the training data but don’t repeat it. It can produce a variety of novel content, such as images, video, music, speech, text, software code and product designs."

Understanding Generative AI can be overwhelming, especially for those unfamiliar with the concept. At the most basic level Generative AI is a subset of machine learning, which involves training models to learn patterns from existing data sets and then use that knowledge to generate new, original content. Not like other AI techniques that are largely focused on recognition and classification, Generative AI models are designed to produce something new for every output requested. These models learn, the more data, the more entries, then the underlying patterns and structures of the training data increase and it uses this understanding to create unique responses, music or even artwork.

Personally, I think the human mind is very creative and complex, it can’t be replaced by this type of tech, but maybe I am naive...

Ok, so what can you do to learn more?

I decided to take one of the many courses on LinkedIn Learning.

It was pretty quick and easy to follow, less than 60 minutes of your time.

So here is what I know, since I started taking an interest in this topic over the last seven months, we have three main types of Generative AI:

Generative Adversarial Networks (GANs): GANs consist of two competing neural networks—a generator and a discriminator. The generator generates new samples, while the discriminator evaluates their authenticity. Through iterative training, GANs can produce highly realistic and diverse outputs, making them popular for image synthesis, video generation, and style transfer.

Variational Autoencoders (VAEs): VAEs are neural networks capable of learning representations of input data and generating new samples based on that learning. They are often used for tasks such as image generation, text-to-image synthesis, and data compression.

Recurrent Neural Networks (RNNs): RNNs are particularly effective in generating sequential data, such as text, music, and speech. By learning patterns from existing sequences, RNNs can generate new sequences that resemble the training data.

These three types of AI models have their strengths and applications in different areas. VAEs are effective at capturing data sets and generating new samples, making them suitable for tasks involving image synthesis and data compression. GANs excel at generating highly realistic content and have been widely used in image and video generation tasks, as well as style transfers. RNNs, with their sequential data processing capabilities, are particularly well-suited for tasks involving text and music generation, speech synthesis, and other tasks where maintaining temporal context is crucial.

Overall, VAEs, GANs, and RNNs represent powerful techniques within the field of machine learning. They demonstrate the diverse ways in which machines generate new content and exhibit creative behaviors. It is truly remarkable at how far this technology has come in the last 5 years. I think by leveraging these functions, we as business leaders can explore options via automation and efficiency tools to continue advancing the capabilities of tech and drive home better user experiences or employee productivity. This will impact your ROI and margins on projects, which is 100% a positive impact, I don't think in the next decade a business will be without Generative AI in some capacity.

As you could expect, there is always another side to the AI coin, as I do have a few concerns that popped up around who owns this data and how do you protect jobs and income levels. Lots of questions around ethics in particular …

In so many of my conversations over the last month, it seems as though a lot of you are predicting marketing and analyst roles being replaced amongst others. Personally, I think the human mind is very creative and complex, it can’t be replaced by this type of tech alone, but maybe I am naive.

It does serve its place in helping perform mundane tasks and jobs, as we can automate so much to allow more productivity, but the question we should start asking is really around security of the data being put into these models. I’ve heard of many examples of people using these AI bots and applications to generate code for programs and new tools.

• Does that not pose a security risk?
• What happens if that model of data is hacked or released?
• How vulnerable are we if there data is leaked?

In my eyes, I see Generative AI has unlocked new possibilities for everyday businesses and leaders, by enabling machines to create, imagine, and generate novel content. From media, educational tutorials and art creation to data augmentation, its applications span across many fields and, transforming industries and challenging traditional approaches and ways of thinking. These tools, powered by a creative human mind, can unlock so many possibilities.

However, responsible development and usage of Generative AI are crucial to address ethical concerns and ensure its positive impact in our lives. As mentioned during the AI learning course above, if you decide to enroll, perhaps we may have a need for ethical boards to direct this tech. We are in control. We can write in policy and adhere to best practices. It is only a reality to think this tech will continue to progress, as Generative AI holds the potential to reshape our interaction with technology and creativity in profound ways, but at what costs to ourselves?

Would love to hear from you, as I recently did a poll on my LinkedIn page, 72% of you stated you were already using AI in your day-to-day at work.

Where are you seeing early success or perhaps any challenges you’ve been faced trying to implement Generative AI at work?

Drop a comment or reach out in a direct message!

#yyctech #yegtech #artificialintelligence #aiforgood #aiethics #machinelearning #smarttechnology

(403) 993-9760

ryanconn@directtec.ca

https://www.directtec.ca/