AI and Human Expertise Hybrid Is The Optimal Solution

Interest in Artificial Intelligence (AI) has skyrocketed. This article consolidates everything you need to know right now, from current developments to how AI is already impacting the financial sector including the AI potential, costs and employment impacts. It concludes AI has massive potential but cannot replace people entirely because human oversight is necessary to ensure the accuracy and ethical use of AI technology. Also, AI cannot replace human creativity and critical thinking skills.

What is Artificial Intelligence?

Artificial Intelligence (AI) is a field of computer science that focuses on creating machines that simulates human intelligence processes to perform tasks that typically require human intelligence. Using algorithms and statistical models, AI analyses data to predict or make decisions. Types of AI include machine learning, natural language processing, and computer vision.

Why is AI Significant?

AI has the potential to revolutionise industries and improve efficiency. Sufficiently trained AI can swiftly analyse vast amounts of data and provide insights that human analysts may not notice - like the JP Morgan example. Repetitive tasks can also be automated and free up human time to focus on complex and creative work, with exceptions like AI image generators. Additionally, AI can potentially solve major global challenges like climate change and disease prevention. Overall, the social and economic impact of AI is expected to be significant and far-reaching.

The AI Race (So Far)

Interest in AI has surged with the meteoric rise of ChatGPT, a Natural Language Processing (NLP) AI tool by OpenAI, which gained more than 1 million users within 5 days of launching in December. Microsoft, OpenAI’s primary investor, applied a similar chatbot to its Bing internet search engine that subsequently challenged Google’s longstanding dominance. Google subsequently responded by publicly releasing Bard, its proprietary AI, 6 weeks later.

ChatGPT, Bing, and Bard are high-profile AI models, but they are not alone. Other companies are also developing AI capabilities, like Bloomberg’s recently announced large-scale language model (LLM) ‘Bloomberg-GPT’, and the AI craze is not limited to companies in the West. Chinese technology giant Alibaba, for example, announced plans to roll out its own AI Chat-GPT style product called ‘Tongyi Qianwen’ with its cloud arm reportedly seeking partners to help accelerate application developments while cutting the price of its cloud service. Sberbank, a Russian financial services firm, also launched GPT rival GigaChat - currently invite-only testing - and Alphabet, Google’s parent company, recently formed a new AI-focused division.

Firms not developing AI capabilities have already begun incorporating AI into their services. JP Morgan, for example, reportedly used a Chat-GPT-based language AI model to analyse 25 years of Fed speeches - other examples include Expedia and Grammarly. With firms already vying to leverage AI capabilities, Emily Dreibelbis, PCMag reporter, demonstrates that each AI model has their respective strengths and weaknesses, depending on how it has been trained and the compute resources available. This raises questions about machine learning (ML) bias as a result of artificial general intelligence (AGI).

The AI Warnings

Despite the recent developments surrounding AI, experts have voiced concerns about the pace of AI evolution. In an interview as early as 2014, the late Stephen Hawking warned about the potential dangers of artificial intelligence which is echoed by Geoffrey Hinton in 2023, who is widely seen as the ‘godfather of artificial intelligence’. 

In line with the warnings, a recent open letter has called for a temporary pause to further AI research to allow better safety guidelines to be developed. At the time of writing, the letter has been signed by 27,655 signatories, including Elon Musk, an initial board member of OpenAI, and Steve Wozniak, Co-Founder of Apple: 

“At some point, it may be important to get independent review before starting to train future systems, and for the most advanced efforts to agree to limit the rate of growth of compute used for creating new models." We agree. That point is now.”

These warnings echo an abandoned 2017 experiment by Facebook, which was shut down after two AI programs began conversing in their own language. Also, in February 2023 Kevin Roose, Technology Columnist at the New York Times, had a lengthy 2-hour conversation with Microsoft’s AI-powered Bing and he concluded that the “AI built into Bing was not ready for human contact”, following unexpected responses like “I want to destroy whatever I want.” Adding to the risks, scientists reportedly cannot explain how AI works.

“An open letter signed by dozens of academics from around the world calls on artificial-intelligence developers to learn more about consciousness, as AI systems become more advanced.” (Kleinman Z. 2023)

Despite the concerns about rapid AI development, world leaders remain bullish on artificial intelligence. Jeremy Hunt, the UK’s Chancellor has reportedly dismissed warnings and Tom Tugendhat, the UK Minister of State for Security urged both the UK and the US to 'stay ahead' and not pause research as China and Russia aim to 'rule the world' by becoming leaders in AI development. Meanwhile, President Biden and Vice President Kamala Harris, who is concerned about the widespread use of AI, have discussed AI regulations with BigTech CEOs.

The Value of AI

The potential applications and benefits of AI are no doubt far-reaching, militarily and economically. While earlier versions of OpenAI’s ChatGPT are available for free, OpenAI has already tested paid versions of ChatGPT (version 3.5 and 4) since January as CEO, Sam Altman, tweeted about back in December (right).

The intent to monetise AI has been expected, especially since the global AI market is estimated to be worth US$1,591.03 billion by 2030 compared to US$119.78 billion in 2022, with a compound annual growth rate (CAGR) of 38.1%. The infrastructure market alone is forecast to be worth US$ 422.55 billion by 2029 with CAGRs between 20.59% and 43.5%. AI is forecasted to potentially increase global GDP by 7% and productivity growth by 1.5% over a decade according to Joseph Briggs, Economist at Goldman Sachs. 

Source: Goldman Sachs (2023)

Financial AI Costs

At a high level, the benefits of AI are impressive but at what cost? This is a multifaceted question from a monetary perspective as Altman’s tweet alluded to, as well as a societal perspective. To address the elephant in the room, OpenAI’s ChatGPT is said to cost up to $700,000 to operate a day and excludes training, which Altman confirmed GPT4 cost more than $100 million.

These costs largely align with Amazon’s CEO, Andy Jassy, who said “Most companies want to use these large language models but the really good ones take billions of dollars to train and many years and most companies don’t want to go through that”. He adds “What they want to do is they want to work off of a foundational model that’s big and great already and then have the ability to customize it for their own purposes.” 

 Developing, training and running AI is undoubtedly a financially costly endeavour. Even without Amazon’s customisable foundation model (Bedrock), firms are already incorporating GPT AI into business operations - for example, JP Morgan, Expedia, and Grammarly - and as AI capabilities mature, it will only accelerate and make workflows more efficient by automating routine operations. 

Other AI Costs

The convenience of automation and machine learning, however, comes at a cost. According to Briggs at Goldman Sachs, AI could replace the equivalent of 300 million full-time jobs, and the phenomenon already seems to be taking hold as Greg Jackson, Chief Executive at Octopus Energy, reportedly said AI was “doing the work of 250 people by answering customer emails” with an 80% satisfaction rate compared to 65% achieved by workers. 

While the potential to cut back on staff exists, Economics Editor at the BBC, Faisal Islam, notes AI LLM are not yet as capable as intelligent adults and Carl Benedikt Frey, future of-work director at the Oxford Martin School, Oxford University argued that while no one can predict exactly how many jobs will be replaced by AI, we can compare it to the introduction of GPS technology and Uber where “the result was lower wages, not fewer drivers. Over the next few years, generative AI is likely to have similar effects on a broader set of creative tasks.”

What Next For AI?

Despite the open letter, AI development is unlikely to be paused universally, but the letter highlighted an urgent need for AI guidelines. Regionally in the US, Biden and Harris have engaged with BigTech CEOs and the EU is reportedly considering “far-reaching legislation” on artificial intelligence.

The EU may be among the first region to implement AI legislation and others will leverage it as a standard to contain the influence of AI - similar to GDPR and local data privacy laws - as Altman states AI “could be used for large-scale disinformation.”

AI Meets Financial Services 

While it is still early days for AI, financial services have already found creative uses for AI. JP Morgan Chase, for example, revealed last month it had “more than 300 AI use cases in production” and follows February’s staff usage restrictions. One of these use cases will likely include investing advice after a fictional investment fund generated by ChatGPT at finder.com outperformed the average of the UK’s 10 most popular funds. 

Jon Ostler, CEO at finder.com, said “It’s not taken the public long to find creative ways of getting ChatGPT to help them in areas where it shouldn’t technically do so. There have been lots of examples of this, notably the person who used reverse psychology to get a list of illegal movie streaming sites, and it won’t be long until large numbers of consumers try to use it for financial gain."

AI Meets Market Data

Finding creative uses for AI has not been limited to the execution side of finance as demonstrated by major market data vendors, like Bloomberg, which unveiled BloombergGPT, a purpose-built 50-billion parameter LLM specifically for the financial industry. Granted, few firms will have the financial might to launch proprietary AI models.

Considering the AI warnings (and potential employment ramifications), I asked GPT whether AI should replace humans to manage market data and IT infrastructure. The response in full: 

“AI has the potential to efficiently manage market data and IT infrastructure. It can process large amounts of data at a faster rate than humans, identify patterns and make predictions. However, it is important to note that AI is not a complete replacement for human intelligence. Human oversight is necessary to ensure the accuracy and ethical use of AI technology. Additionally, AI cannot replace the creativity and critical thinking skills that humans possess. Therefore, a combination of AI and human expertise is the optimal solution for managing market data and IT infrastructure.”

There will undoubtedly be additional AI-centric insights, which will explore in greater detail, but based on the hybrid recommendation I queried how AI can help and it responded with the below.

“AI can help manage market data and its IT infrastructure by automating data analysis, identifying patterns and trends, and providing real-time insights. With AI, businesses can streamline their data management processes, reduce errors, and make more informed decisions. Additionally, AI can help optimize IT infrastructure by identifying potential issues and providing proactive solutions. This can lead to improved performance, increased efficiency, and reduced downtime. Overall, AI has the potential to revolutionize the way businesses manage their market data and IT infrastructure.”