Generative AI 101- Whats the buzz all about?

Generative AI is the new buzz around Silicon Valley circles nowadays. Even those that felt tentative about the timing of generative AI, were forced to concede when Jasper AI became at unicorn this October, raising $125 million at a $ 1.5 billion valuation in their Series A!

But what exactly is Generative AI and is it really that buzzworthy?

So we thought we'd begin with a crisp synthesis around this new phenom, and why it must be noted

What is generative AI and why should you care?

Generative artificial intelligence is a subfield of AI that deals with the generation of new content, such as text, images, or ideas. That is, the machine generates some thing new, rather than analysing something which already exists.

Earlier, with AI you could take an image and predict if the image is a or b, human or animal. As the machine got more data to “learn” from it could identify male or female. Similarly, for text, AI models became better at sentiment analysis.

Put another way, other types of AI, such as predictive AI or prescriptive AI, focus on analyzing existing data in order to make predictions or recommendations. But generative AI takes things one step further by actually creating new data.

In an earlier post I had referenced with an example how creative tasks can be scaled using generative AI applications. Here's a refresher -

Why now?

Prior to 2015, AI models had started to become widely used for analytical and predictive use cases. For instance, the YouTube algorithm since became better and better at showing you more relevant content every time you opened the site. But we couldnt expect machines to generate human like code or creative tasks back then for multiple reasons

Attention is all you need

Post 2015, Google's paper- Attention is all you need, spoke about- how to create more effective AI models by using a technique called "attention." This involves giving each individual data point its own level of attention, which allows the AI model to focus on the most important information. By doing this, the AI model can learn and remember information more effectively.

The paper introduces a new technique called "attentional learning", which involves giving machines more attentional control. This allows them to better focus on specific parts of the data they are learning from, which leads to improved accuracy

There are several reasons why it is easier to run generative AI models in 2022 than in 2015-

1. For one, the development of new algorithms and software has made it easier to train generative AI models.

2. In addition, the rise of cloud computing has made it easier to access the resources needed to run these models.

3. And finally, the increasing availability of data has allowed for more accurate and efficient models

Who are the top 5 notable generative AI startups, you should know about?

1. OpenAI : OpenAI is one of the leading lights in generative AI. The company was founded by several heavy hitters in the tech industry, including SpaceX co-founder Elon Musk. OpenAI’s goal is to “advance digital intelligence in the way that is most likely to benefit humanity as a whole.” To this end, the company is researching a number of different areas in generative AI, including natural language processing and robotics.

2. Google Brain : Google Brain is the search giant’s in-house AI research team. The team is responsible for some of Google’s biggest AI achievements, including the development of the RankBrain algorithm. Google Brain is also working on a number of other interesting projects, such as using AI to improve the usability of Google products and services.

3. DeepMind : DeepMind is another major player in the world of AI. The company was acquired by Google in 2014 for a whopping $500 million. DeepMind is perhaps best known for creating the AlphaGo AI, which beat a world champion at the game of Go.

4. Neurala : Neurala is a startup that is focused on making AI more accessible to businesses. The company’s goal is to “enable a new generation of smart products that see, feel, and think like humans.” Neurala’s technology is being used in a number of different industries, including retail, automotive, and drones.

5. Casetext: Casetext is a startup that is using artificial intelligence to revolutionize the legal research industry. The company’s goal is to “make the law more accessible and understandable for everyone.” Casetext’s AI platform is used by lawyers and law students all over the world.

Generative AI refers to the ability of computers to create new data that is similar to existing data. This has enormous implications for a wide range of industries, including healthcare, finance, manufacturing, and retail. Here are just a few examples:

How will generative AI change the world as we know it

As we now know, generative AI refers to the ability of computers to create new data that is similar to existing data. This has enormous implications for a wide range of industries, including healthcare, finance, manufacturing, and retail. Here are just a few examples:

1. Healthcare: Generative AI can be used to create 3D models of human organs for medical research and training. It can also be used to develop new treatments for diseases by creating virtual patients who “test” potential therapies.

2. Finance: Banks can use generative AI to create detailed risk profiles for individual customers. This information can then be used to make more accurate decisions about lending and investment.

3. Manufacturing: Generative AI can be used in conjunction with 3D printing to create customized products on demand. This could revolutionize the way products are manufactured and reduce the need for inventory storage.

4. Retail: Online retailers can use generative AI to design custom product pages that match the style and tone of their website. They can also use it to create realistic product images that show customers what the product will look like in use.

Conclusion

The key opportunity here is newness of GAI itself. Since it's so new, it's an open field for young fast, technical teams which will just need to spend 8-9 months to educate themselves and come out ahead. Since underlying AI models are evolving so quickly, any technical team that is nimble and can iterate fast, upon getting their finger on the pulse of network effects can expand very quickly - Jasper (mentioned at the beginning of this blog) is a shinning example of that.

We are excited to witness and hear from startups coming out of this space in India and the Bay area. If youre someone building a startup in this space, or know someone who is, we would love to hear from you all!