Highlights of recent developments in the field of generative artificial intelligence

Generative AI was the explosive breakthrough tech trend of 2023, catching the interest of not only techies and futurists but popular audiences, too, with its capacity to generate seemingly human-like material.

With tech giants including Microsoft, Google and Meta pouring their apparently limitless resources into the technology, we can expect equally game-changing innovation in 2024.Generative AI is an incredibly powerful tool that will play an important role in democratizing access to the transformative potential of AI. And I believe everyone has to be informed of what's coming in order to properly examine the influence it will have on our lives. So here’s my summary of what I think to be the primary themes, spanning new technology advancements as well as societal difficulties we’re going to confront.

Bigger and more powerful models

Generative AI applications are impressive due to their training on massive datasets. GPT-4, the engine behind ChatGPT, is reportedly trained on over one trillion parameters. Other large language models (LLMs) like Google’s PaLM2 and DeepMind’s Gopher are trained on hundreds of billions of parameters. There are rumors circulating about the potential release of GPT-5, and it is anticipated to be another significant advancement in terms of size. While increasing the size of the dataset is not the only method to enhance an LLM, it has proven to be the most reliable way to make improvements. As a result, I anticipate the development of new generative AI models to continue in this direction in 2024.

Generative Design

Generative design is becoming increasingly popular in the field of designing physical products and services. Tools under this category enable designers to input specific requirements and available materials in order to receive blueprints and instructions. Platforms like Autodesk are integrating generative AI features, empowering product designers to quickly develop numerous prototypes as digital twins for simultaneous testing, leading to the creation of more durable, efficient, and environmentally-friendly product designs.

Generative Video

The younger generations prefer video as their media format of choice, so it's no wonder that AI toolmakers have been quick to develop tools that cater to this trend. Creating video content has historically been costly, largely because of the need to hire a skilled human professional to create, shoot, or edit footage. However, with generative video tools, anyone will be able to produce high-quality video content at a fraction of the cost by 2024.

Generative Audio And Speech

In 2024, AI-generated audio and speech are expected to become as common as AI-generated words and pictures have become in 2023. I anticipate that the slightly robotic characteristics that are often noticeable in most AI-generated speech will begin to disappear as the tools improve in simulating the nuances and intonations of human speech.

Multi-Modal Models

Generative AI tools, algorithms, and LLMs typically focus on simulating a single mode of expression, such as language, visuals, or sounds. However, there is a shift towards "multi-modal" generative AI. The upcoming version of OpenAI's ChatGPT is expected to have the ability to comprehend and analyze images, as well as respond to voice commands. Meta has also showcased a model that can integrate images, text, audio, depth, and inertial data. This trend is projected to become more prevalent in the coming year, making it commonplace to engage with AI about images or videos in the same manner as we do with text today.

Prompt Engineers In High Demand

Programming engineers train generative AI systems by outlining their objectives using the most effective terms. This role has been dubbed as "the hottest new job in tech" for 2023, and the demand for it is expected to rise in 2024. People have also referred to this job as "AI whispering," emphasizing the skill of bringing out the best in systems that may sometimes appear intricate, incomprehensible, and daunting.

Autonomous Generative AI

Autonomous Agents refer to a category of generative AI applications that can independently function by continuously generating and responding to prompts. This allows them to perform more complex tasks compared to chatbot-style generative agents, which only react to prompts and then await further instructions from users. An example of an autonomous agent that gained recognition in 2023 is AutoGPT. In 2024, I anticipate advancements in this area as AI developers strive for a "generalized" AI capable of adapting to various tasks assigned to it.

Generative AI-Augmented Apps And Services

By 2024, 40 percent of enterprise applications will come with conversational AI as an embedded feature, according to AIM Research. We see it happening already with companies like Microsoft and Adobe eagerly incorporating chatbots and natural language interfaces into their platforms. Even popular social messaging app Snapchat has introduced a generative AI bot to users' contact lists, allowing them to ask questions or simply interact with a virtual friend. In the near future, it's expected that app developers will prioritize adding chat interfaces to enhance user engagement and overall customer experience.

Generative AI in Schools And Education

During the last century, there was a lot of resistance to the introduction of electronic calculators in classrooms. Traditionalists were concerned that students would become too reliant on them and lose the ability to do basic calculations without assistance. However, what actually happened was quite the opposite. Students were able to tackle more advanced calculations at a younger age. Eventually, calculators became so widespread that the question of whether we could function without them became irrelevant. Similarly, we can expect to see debates about the use of generative AI by students in 2024. At the same time, teachers and professors will become accustomed to using it to create notes, reports, lesson plans, and course summaries.