We don’t think it’s hyperbole to say that in the past few months, the world has experienced a paradigm shift in artificial intelligence technology.
As a company that has been developing conversational AI since 2019, we couldn’t have dreamed up a better time to show the world our next exciting leap forward.
That’s why we’re eager to introduce Daizy AI 4: the world’s most advanced investment chatbot.
We believe Daizy AI Chat is the tool that will empower the humans who will help the next generation of conscious investors.
If you didn’t click the link above, just know Daizy AI is absolutely free at daizy.com. We know this is just the beginning of the blog today, but if you want to leave now and check out Daizy, please do.
However, if you want to learn a bit more about what’s new in our fourth major release, how we got here, and our thoughts on navigating this newly accessible world of conversational AI, stick around!
How did we get here?
As you can imagine, building a cutting-edge investment chat is no small task.
That’s why we’ve been constantly adding, subtracting, changing, iterating, tweaking, re-evaluating, and everything in between for the past 4+ years.
As we’ve watched OpenAI and other AI companies rise to the spotlight over the past few months, we couldn’t be happier to watch the world realize together the power and potential of generative AI. This is the same realization we had when we started building Daizy - it’s what inspired us then and continues to drive us forward today.
A look into of the Daizy v4 experience
We also recognize that with great power comes great responsibility.
It is going to take mindfulness, trust, patience, and persistence for each of us as individuals and all of us together to leverage this technology for its best. This can be exceedingly difficult when it feels like the AI world is moving a million miles an hour. It’s easy for FOMO (fear of missing out) or FOFB (fear of falling behind) to take over.
So let’s discuss how to navigate this brand-new world of AI as wisely and effectively as possible.
First, what exactly are all these GPTs?
Because the whole world is talking about one form or another of OpenAI’s GPTs (ChatGPT, GPT-3, GPT-4), and because we’ve integrated GPT and other models into Daizy, we figured it would be worthwhile to explain a little bit!
GPT stands for Generative Pre-trained Transformer and is a highly advanced language model that is useful for a wide range of language-related tasks.
It is trained on a massive amount of text data and can generate high-quality natural language responses when given any number of a wide range of tasks. It is particularly useful for applications that require language understanding and generation, such as language translation, content creation, and chatbot development. Some of its notable features and use cases include:
But more importantly, how do we use GPT?
In order to produce the Daizy experience - the ability to provide real-time, conversational, and curated investing insights - we employ several different large language models for use in every query.
Every input you enter into Daizy goes through a series of engines and language processors, and GPT is just one piece of the puzzle. Each piece of AI we use in the Daizy ecosystem is specialized for certain tasks: some are great at understanding conversational English, some are great at parsing and identifying data, and some are great calculation engines.
Right now, we use GPT on the back end of our data crunching to help fill in the gaps when someone asks definitional questions about financial terms or wants to summarize data. We use GPT like an aftermarket car upgrade - we had a functioning top-of-the-line car before, but now it’s just a little bit fancier.
We’re also not tied to GPT. We experiment with many alternative models and use them when they are better at a specific task.
We continue to explore using GPT on the front end as well. Translating a person’s writing into a display of complex data involving millions of calculations is a difficult computing problem. Natural language and computation don’t always mix well.
GPT is unable to access real-time investing data, and it’s not great at even making simple numeric calculations, but it does happen to be pretty excellent at translating natural language into inputs that other AI’s understand. In the future, we will likely use GPT to help the rest of the Daizy ecosystem better understand each individual user’s input and then generate the most helpful answer.
Guard-rails: “GPT may be 90% correct but Financial Services needs to strive for as close to 100% as possible”
Since the release of ChatGPT and the ability for companies to use OpenAI’s APIs, suddenly there are thousands of new “AI companies” sprouting up left and right. Many of these are just GPT wrapped in a fancy user interface.
It’s important to distinguish between the companies that have developed and trained their own AI models and the ones that claim to be unique but are just copies of another open-source model.
As mentioned above, Daizy has built its own AI tools, and is enhanced with an ecosystem of different AI technologies. Some are proprietary that our team has been developing since 2019, and some are third-party partners. The newest release of Daizy is the fourth major version, and we’re not slowing down anytime soon.
We use each AI tool very specifically, including GPT. We have and will continue to work very actively to keep GPT on a tight leash within Daizy, so that the answers it provides only relate to finance and investing, use and label institutional grade quality data sources, and so that it is never called upon to perform a function it wasn’t intended or worse hallucinate. Hallucinations are not welcome here.
Source: technopedia
As we all learn to use these tools and implement them into our daily lives, we believe there is an important and necessary function of AI companies to be transparent about how they are using AI. Proper governance is essential to building trust in these technologies - otherwise, the risks outweigh the potential.
Risks and Rewards
The public discourse around AI is fairly split between enthusiasts and critics.
The enthusiasts are ravenously searching for ways to integrate AI into their work and lives. While we love the passion - these conversations often lack a depth of understanding and don’t distinguish between the strengths and weaknesses of the tools used.
The critics, on the other hand, often throw the baby out with the bath water. Just because certain AI tools make frequent errors while performing certain tasks, doesn’t mean that all AI is useless and subpar.
We’d like to find ourselves somewhere in the middle - mindfully enthusiastic, striving for the best AI can deliver while constructively critical to manage the downside risks. We can’t let the extreme sides of the narrative isolate us because, ultimately, AI is built by people for people - for all people. It is only useful to the extent to which we participate in creating and guiding the technology, and subsequently the extent to which it effectively serves all of us. Without humans, AI has no future and no purpose.
So how do we set ourselves up for success in this new paradigm? Here are three key principles we stand by as we continue to embrace these technologies with a sense of humanity and balance.
Ownership
What do you want your relationship with AI to be?
Even though we’re talking bleeding-edge tech here, it still all comes back to people. People are going to direct the course of this technology for better or worse.
We want to empower people to benefit from AI in the wealth space, and to build trust via sustained engagement - not writing it off or treating it like a fad.
We should seek to take ownership, not blame the tool. Learning the capabilities, nuances, and weaknesses of different types of AI models is absolutely necessary to effectively and safely integrate them into our world. Much of this comes down to each individuals’ perspective and actions.
Data Integrity
If people are one success factor, data is another. AI’s live or die by the data they’ve been trained on.
Many language models have been crucified on the public stage because of failing simple calculations, not being able to crawl the web, or “making things up.” Part of this problem goes back to the first point - the tool is being misused, but the other part begs the question “Where are all of these instantaneous outputs coming from?”
An important criticism of many language models is that right now many of them are not or are unable to cite their sources. And just like you couldn’t get away with lack of citations in high school, no matter how brilliant the work, we’re all now realizing that generative AI shouldn’t be able to either.
Many AI companies, including us, are rapidly advancing toward being able to cite the sources of the outputs generated. We believe this is absolutely crucial to the viability of language models and generative AI going forward.
Purpose
Lastly, we have to fit and train AI for purpose. The best AI companies will have specific tools built and trained for specific purposes.
Our purpose is creating a future of conscious investing by providing AI to help the advisors working with the next generation investing knowledge and insight through a tool that is accessible, best-in-class, and helpful/compatible with humans.
Stepping into that purpose means sacrificing a lot of other intriguing missions. We won’t be the best at AI art, writing, or code generation, but we believe we are the best AI tool for wealth such as including answering your finance and investing questions.
What’s Next
Daizy AI 4 is live, accessible anytime, and free for anyone with internet access. We are so excited to share it with the world that we made it our homepage.
Soon, we’ll be integrating the ability to ask Daizy questions about your real, live portfolio. If you already have a Daizy account, your investing profile will be integrated into Daizy and you’ll automatically be able to get on demand insights into your investments. So stay tuned.
Daizy is a collaborative creation built for all of us. If you have ideas, feedback, or want to partner with us, please reach out!
As we continue to build the most advanced conversational AI for investing, we invite you to join us. After 4 years on this mission, we’re still at the beginning, and the more we build together, the better the technology will serve our needs and better all of our lives.
