“Investing in machine learning stocks can be a lucrative opportunity for investors looking to capitalize on the growth of this rapidly-evolving technology. Machine learning, a type of artificial intelligence (AI) that allows computers to learn and adapt without being explicitly programmed, has numerous applications across a wide range of industries, including healthcare, finance, and retail. As such, there are many companies that are actively developing and utilizing machine learning technology and could potentially be good investments for those interested in this space.
“One way to invest in machine learning is to look for companies that are actively developing and selling machine learning products or services. These could include companies that specialize in machine learning software, hardware, or consulting services. For example, NVIDIA is a well-known company in the machine learning space that designs and manufactures specialized graphics processing units (GPUs) that are used for machine learning and other high-performance computing tasks. Other companies in this space include IBM, Microsoft, and Google, which all have extensive machine learning capabilities and offer a range of products and services related to the technology.
“Another approach to investing in machine learning is to look for companies that are using machine learning to drive innovation and efficiency within their own operations. For example, some retailers are using machine learning to optimize their supply chain and inventory management, while healthcare companies are using it to improve patient care and diagnoses. Investing in these companies could provide exposure to the growth of machine learning as a technology, as well as the potential benefits it can bring to these industries.
“It’s important to keep in mind that investing in machine learning carries some risk, as with any investment. For example, the field of machine learning is constantly evolving, and there is the potential for companies to become outdated or lose their competitive edge if they are not able to keep up with the latest developments. Additionally, there may be regulatory hurdles or other challenges that could impact the adoption and growth of machine learning technologies.
“One way to mitigate some of these risks is to diversify your portfolio by investing in a range of machine learning companies or industries that are using the technology. This can help to spread the risk and potentially smooth out any potential ups and downs in individual companies or industries.
“Overall, investing in machine learning can be a potentially lucrative opportunity for investors, but it’s important to do your research and consider the potential risks as well as the potential rewards. It may be helpful to consult with a financial advisor or professional to get a better understanding of the risks and opportunities involved in investing in machine learning stocks.”
Not bad, right? The entry above was actually generated by OpenAI’s ChatGPT bot. If you’re interested in the development of machine learning I highly recommend signing up for personal access.
That was generated by the large language model (LLM) responding to a prompt asking it to “write 400 words about investing in machine learning stocks.” It delivered that response in approximately 20 seconds, for free, and came in at 447 words. Other than the requirement that anything produced by the bot is flagged as AI-generated, that’s not much different content-wise than anything I would expect to read from a half-decent copywriter, and you can’t beat the price.
[text_ad]
There are a few places in that text where I would have made some edits had it come from a Cabot analyst, but for the most part, it’s clean, competent (but not great) copy. But it’s perfectly serviceable copy if you’re not in the business of the written word.
A Case for Machine-Learning Stocks
If you, for instance, ran a bed and breakfast in Vermont and wanted to have a page on your website that’s 1,000 words about the history of Vermont, ChatGPT could handily deliver. Or, if you’re looking to raise capital to run that bed and breakfast, ChatGPT could even write you a handy business plan … like this one …
“Our bed and breakfast, located in the beautiful state of Vermont, will offer guests a cozy and comfortable retreat in a stunning natural setting. We will offer a range of well-appointed guest rooms, each with its own unique charm and character, as well as a delicious daily breakfast made with locally-sourced ingredients. In addition to the comfortable accommodations, we will also provide guests with access to a range of amenities and activities, such as hiking trails, skiing, and local sightseeing opportunities. Our friendly and attentive staff will be on hand to ensure that guests have a relaxing and enjoyable stay.”
As you can probably guess, that was also AI-generated, this time in response to a prompt to “write a 1-paragraph business plan for running a bed and breakfast in Vermont.”
The reason I’ve led with those two examples is to show you, first-hand, what these machine-learning models do well, and what they don’t. What they do well is look at their datasets and predict the next most logical word in a sequence of words that will make their output look as close as possible to “human.”
Whether something looks “human” or not is dictated by the training mechanism, and can be fully automated, automated with human input, or entirely dictated by human input (incredibly time-consuming). Since the output is largely a context-specific (provided by the prompt) guess at what word logically comes next, it’s essentially as close to “average” writing as you can get.
That makes it useful for summarization (ChatGPT recently summarized a book that my daughter was reading for me so I could help her with a book report), or even some administrative tasks (see the business plan above).
There’s an ongoing mission at Alphabet (GOOGL) to turn their search functionality into the Star Trek computer. If you haven’t seen it, users simply ask the computer a question and receive a factual response. They don’t have to click on a website that maybe has the best information, they just receive an answer to their question.
ChatGPT is as close to that result as anything I’ve seen before, and I’ve actually started using their open beta for some things that I would normally use Google search for (asking for a recipe good for picky eaters) because I avoid some of the pitfalls of search (reading 10,000 words about the beauty of Vermont before finally getting to a recipe for cheesy chicken and rice casserole, for instance).
Weaknesses of Machine Learning
Where these models tend to disappoint, however, are specificity and creativity. As you can see in the image at the top of this page (created by AI image generator NightCafe based on the prompt: “a digital image of someone investing in machine-learning stocks”), these AI models are producing content without actually creating anything.
The initial outputs typically look like you’d expect from software “doing its best” to replicate whatever prompt you provide, but without knowing what it’s producing, it pretty quickly enters the uncanny valley.
Prompters often iterate their images through different versions, and you can get some truly remarkable artwork, like the image that won first place in the digitally manipulated photograph category at the Colorado state fair. But aside from the creativity of the prompter and the direction that his or her inputs send the model, the model is just averaging the creativity of things that already exist.
When you start getting into specifics, you can also run into trouble. ChatGPT’s coding output has been banned from Stack Overflow because “the answers which ChatGPT produces have a high rate of being incorrect, they typically look like they might be good and the answers are very easy to produce,” as one site moderator explained.
Like with the image, the specific code is an approximation of what the training model expects us to expect to see. It’s broadly “right” until it comes under scrutiny.
To put it another way, you could decide tomorrow morning to open a bed and breakfast in Vermont, get an expansive business plan, a website full of content, and even tips on how to interact with your guests using nothing but ChatGPT. And you could get all that done before lunchtime. But I wouldn’t trust it to come up with the next killer screenplay.
As we’ve seen though, especially over the last three months, these technologies are developing very rapidly, and even once the novelty wears off, there are plenty of viable uses that could make models like this into what is essentially a no-hassle, machine-curated Google search.
Once these scale up, there’s also the question of monetization. Most of the image generators require you to pay for credits, and ChatGPT and other large language models are currently being subsidized by their developers (ChatGPT is estimated to cost roughly $100,000 per day).
How to Invest in Machine-Learning Stocks
DeepMind is Alphabet’s wholly-owned subsidiary. It turned a profit for the first time last year and spun its drug discovery division off into a separate unit. Alphabet has the tech, capital, and know-how to remain a leader in the space, but as with most tech, one breakthrough can make all the difference.
Microsoft (MSFT), on the other hand, currently has a $1 billion+ investment in OpenAI (which is largely venture-funded), the company behind ChatGPT, although specifics of the deal are not disclosed. But after spending a few weeks with ChatGPT it’s not hard to see the pathway for an improved Bing search experience.
Both GOOGL and MSFT offer indirect investment in machine learning tech, as do Nvidia (NVDA) and IBM (IBM), as ChatGPT’s introductory blurb highlighted. But the fates of these stocks will likely be dictated more by the ongoing bear market in growth stocks than any progress in machine learning. That said, if you had to bet on the “next Google” for the AI revolution, odds are it’s probably “still Google,” or one of the other three tech giants.
[author_ad]
