While the banking crisis may have interrupted funding flows for a limited period, the hype cycle in generative AI has momentum with staying power.
Deep Dive Into Generative AI and What Will Drive Tomorrow
By Louis Lehot, partner and business lawyer at Foley & Lardner LLP
Despite the banking crisis that has been shaking up Silicon Valley this March, the hype cycle in generative AI continues unabated. There is a growing consumer and enterprise appetite for the technology, interest in which has exploded since the November 2022 release of the ChatGPT application by OpenAI, combined with a growing shortage of knowledge workers, that could result in substantial unrealized revenue. As a result, more and more companies are jumping on board. And with all this buzz, there is an even greater interest on the part of investors, who are looking to position themselves ahead of each new wave of development. Meanwhile, consumer advocates and regulators are pushing for increased protection of intellectual property and personal privacy.
In this post, we explore how we get to this point, why the time for generative AI is now, the impact on the entrepreneurial system, where dollars are being deployed, and where are we going next.
How did we get here?
Artificial intelligence is the development of computer systems performing tasks that typically require human intelligence; for example, visual perception, speech recognition, decision-making, and translation between languages. It’s a combination of computer science as applied to data sets to enable problem-solving. Generative AI is an algorithm that can create new content, including audio, code, images, text, simulations, and video. The key to generative AI is extensive language modeling that can analyze and learn from natural language interfaces, partnered with media that can procure natural language inputs, with AI core foundation modeling and infrastructure that can analyze inputs to create generative outputs.
The phrase was coined in December 1997 when computer scientists Sepp Hochreiter and Jurgen Schmidhuber invented long short-term memory networks, which improve memory capacity in neural networks, allowing for pattern recognition in training data. In 2012, Alex Krizhevsky gave us AlexNet, a convolutional neural network trained on graphical processing units, breaking 75% accuracy in identifying images from a manually tagged database. While Google has led research and development in the field for more than a decade, focused on a product called “BERT,” OpenAI formed and released its first generative pretrained transformer in June 2018 called GPT, ushering in a new era of LLMs. A year later, Microsoft invested $1 billion into OpenAI, launching the modern-day AI arms race, culminating in a new text-to-image model called Dall-E, and ChatGPT, released in late 2022, which has been adopted at warp speed.
Why now?
According to one of the world’s leading technology investors and General Catalyst managing director Deep Nishar, ChatGPT has “fired up the imaginations of nontechnical people. It’s probably the fastest thing that ever got 100 million users using it all at once.” With ChatGPT, AI stopped being the tech of the future and became the tech of the present. It has fundamentally changed people’s perception of the future of work, and in the enterprise, it has gone from a novelty to a must have. Enterprises are redeploying research and development dollars to AI, sales cycles are shortening, and widespread availability of large language models has decreased the delivery cycles.
Nishar points to four key developments that have propelled AI forward:
- The proliferation of algorithms;
- The vast amounts of data that are connected to algorithms from the internet, including text, speech, image and video;
- The rise of computing power; and
- A new generation of computer programmers who understand the math in the algorithms.
What is the impact to the entrepreneurial ecosystem?
New companies are being formed and financed to attack the many use cases. At the same time, Big Tech companies such as NVIDIA, Google, Microsoft, Meta and even government agencies are leading an arms race and are locked in battle to innovate and deploy new AI technologies as layers on top of existing products or in place of human workers, each deploying billions of dollars to win. Microsoft announced a multiyear, multibillion-dollar investment in OpenAI and exclusive integration with its Azure system in January. Stability.ai and Amazon have partnered on AWS. Google announced BARD and invested $300 million in Anthropic. General Catalyst led a $350 million round into Adept AI.
Where are dollars being deployed?
Natural-language user interface (NLUI or LUI) is a type of computer-human interface where the user and the system communicate using natural, human language. So users are communicating with a computer using their spoken language. Most of us use this daily on our phones or other devices (Siri, Alexa, etc.). The largest areas of investment within this specific segment were chatbots, voicebots, and personal assistants, which captured $544.9 million in 2022 (59.6% of all dollars invested), according to PitchBook Data.
2D media is exactly as it sounds — any artwork that exits in two dimensions, such as paintings, drawings or prints. AI can be used not only to create 2D media but also to convert 2D media into 3D media. In this area, investors seem to see the most possibility in avatars, video generation and editing, grabbing 37.7% and 40.8%, respectively, of total dollars invested.
AI Core and biotech also brought in impressive numbers. PitchBook found AI Core, which includes foundation model developers and infrastructure for model development, raised $5 billion between 2018 and 2022. Biotech startups using generative techniques have also piqued the interest of venture capital investors, with $1.6 billion invested during the same period.
In light of the promise of generative AI, enterprise applications are racing to integrate the capabilities. Natural language interfaces will likely be the primary catalyst of this growth. In fact, PitchBook in a recent report said that it “expects the market at a 32.0% CAGR to reach $98.1 billion by 2026.”
Where we are going next?
As we look forward, the costs of foundation model training are dropping. Custom hardware and accelerated software tools are making it cheaper to train new LLMs. Price points are accessible to startups. Meanwhile, chief information officers at larger corporations are pushing for digital transformation and AI adoption across the enterprise. According to a report published by the Massachusetts Institute of Technology in September 2022, only information technology, supply chain and finance departments were gaining widespread adoption of AI, and only at the 40% level. We believe the launch of ChatGPT4 is driving adoption deeper into those functions and driving expansion into sales, marketing, product development and human resources.
ChatGPT has gone viral, and end users and consumers are adopting it. While yesterday’s estimates of the market size and opportunity for AI software looked impossible to attain, it now looks understated.
This space has fantastic potential, and it will be interesting to see how it ultimately impacts industries across the board and becomes an even more significant part of our daily lives. One area that will be critical to watch is how lawmakers approach regulation and what kind of implications those regulations will have on this rapidly evolving technology. Although, as with most technology, regulation needs to catch up to advancement.
What are the guard rails of intellectual property? Who owns the content output if the inputs were copyrighted? Who is responsible for copyright infringement? What if the output is wrong and it drives tortious conduct? Who will bear the risk of liability? Will it be insured? How can you protect your image, likeness, speech and images, and how can you enforce them?
It is a brave new world; we may only be in the early innings of a new technology revolution. For startups looking for funding, we believe the keys to success will lie in foundational technology combined with unique algorithms that have some demonstrated utility and traction in at least one vertical but with applications across verticals. While the banking crisis may have interrupted funding flows for a limited period, the hype cycle in generative AI has momentum with staying power.
Louis Lehot is a partner and business lawyer with Foley & Lardner, based in the firm’s Silicon Valley, San Francisco, and Los Angeles offices, where he is a member of the private equity and venture capital, mergers and acquisitions, and transactions practices, as well as the technology team.
Originally Published here.
