The Changing Focus on AI in Our Minds and Business

There has been a lot of excitement and hype around Generative AI (GenAI) lately. ChatGPT, Midjourney, Bard and other systems have captured the public's imagination with their ability to generate human-like text, images, and more. Some proclaim these technologies will usher in an AI revolution, transforming how we work, create, and live. But beneath all the AI euphoria lies a more complicated reality.

Behavioural Changes

I recently ran an informal survey to see how much GenAI has been adopted. In the survey, I asked,?

"In your day-to-day work, when you encounter a problem or need an answer, where do you go to get a solution? "?

While the sample size was small, I would not be surprised if the?results ?reflected a broader trend. Remarkably, 19% of respondents said they now turn to Generative AI as a go-to resource. Ten months ago, this would not have been an option for most people. Nearly one-fifth of respondents have incorporated Generative AI tools into their daily workflow for solving problems and getting answers in?less than a year.

GenAIs rapid adoption highlights how quickly these technologies have begun permeating many professional domains. It also foreshadows AI's potential to alter how knowledge work is accomplished. We are seeing a significant change in focus as GenAI increasingly takes its space in all our minds.?

Textbook case study: How to launch a product

OpenAI, the creators of ChatGPT, nailed their launch strategy. They were the first GenAI globally used, and they introduced AI to the general public. At the same time, OpenAI's CEO Sam Altman generated?positive press ?by explicitly asking governments to regulate AIs, leading to a media storm as almost every big name in the tech industry signed a petition to?Pause Giant AI Experiments .?

The free marketing campaign worked as ChatGPT currently has?over 100 million users , and the website generated?1.6 billion visits in June 2023 . In rapid succession, openAI released development tooling enabling developers to begin playing and embedding Generative AI into an ever-increasing array of services.

Rarely are the first movers the winners. This case might be the exception.

The impact on the boardroom

While technical and frontline teams are closest to AI implementations, these technologies also have profound implications for senior leadership. The strategic opportunities and threats posed by AI are becoming boardroom-level conversations.

Management teams are?increasingly focused on AI opportunities ?on company earnings calls, and more mentions of AI predict higher capex dedicated to it. No doubt I do not need to tell you this. Just look at all the headspace it is taking in your company. It's a safe guess that most readers of this article are part of an organisation that has or is integrating an "AI" solution (probably chat) into their roadmaps, touting that this will revolutionise their product.

To capitalise on AI's potential, boards must take an active governance role. Directors must ask questions about their company's data and analytics capabilities required to fuel AI. They should understand how AI may enhance products and services in the future or pose a disruptive threat. Boards must also consider second-order impacts on their brand, culture, regulations, and talent.

The Hidden Gem of Enterprise AI Applications

While much of the hype around AI focuses on underlying infrastructure and tools, the most significant impact will come from enterprise applications. Though less glamorous, it is at the application layer where AI can provide immense business value by enhancing workflows, providing insights, and ensuring a product satisfies its customer's needs.

The key is identifying high-impact user cases that align with strategic goals. Then, developing AI solutions purpose-built for those needs. With suitable applications, AI can automate processes, uncover hidden insights, improve decision-making, and delight customers.?

In my opinion, this is where OpenAI will target its ChatGPT Enterprise offering. ChatGPT and other large language models show how AI could transform businesses through enterprise applications. One intriguing possibility is a secure, sandboxed environment where companies can host structured and unstructured data. Then, leverage the power of GenAI to produce solutions.

This could include website logs, sales transactions, inventory, communications (email/slack/intranet), contracts, legal documents, etc.?

We currently need teams of data engineers to design, configure and execute?ELT/ETL ?pipelines, analysts, and subject matter experts to collect, process, contextualise, and extract value from our data.

With enterprise AI platforms, knowledge workers simply need to write prompts to query the data and receive automatically generated reports, insights, and recommendations. This has the potential to automate vast amounts of manual analysis work.

For example, A customer service rep could write a prompt that analyses customer support logs to identify common complaints and suggest process improvements to the value chain to satisfy specific customer needs.?

Another example could be an account manager writing a prompt that synthesises data from access logs, product usages, and behavioural trend data to predict churn risk for specific customer segments. The possibilities are endless. This directly puts enterprise AI products in a very lucrative competitive space occupied by vertexAI and Snowflake .

The Impact of AI on the IT Industry

A recent?Goldman Sach ?report on the potential effects of AI stated:

"Using data on occupational tasks in both the US and Europe, we find that roughly two-thirds of current jobs are exposed to some degree of AI automation, and that generative AI could substitute up to one-fourth of current work. Extrapolating our estimates globally suggests that generative AI could expose the equivalent of 300m full-time jobs to automation."

The report shows the potential industries AI will disrupt with an estimated percentage. It's fascinating reading which trends in the right direction when we look at its impact in just under a year.

Third in this list is "Architecture and Engineering". We can empirically see this disruption in the decline in Q&A sites like Stack Overflow, which have witnessed year-over-year traffic decreases since the release of ChatGPT 3.5, contributing to over a 50% decrease in new visitors this July compared to July 2022 .

Applying the Goldman Sach estimate to today's IT sector, AI automation could reduce between $409 million to $2.04 trillion from corporate wage bills annually (1). Even considering just the IT industry, these projections are staggering.

Tipping point: Next Generation AI, our need for safeguards & Regulation

Our unspoken fear may be around AI's potential in behavioural prediction - if systems can accurately model human psychology and personalities, that data could be abused for targeted manipulation and persuasion.

If AI masters the art of behavioural prediction, there are potentially both positive and negative outcomes. On the positive side, it could allow for more personalised education, healthcare, marketing, and experiences that cater to our preferences and needs. However, it also raises dystopian concerns around mass persuasion and manipulation if such data is misused. Addictive social media and targeted disinformation could become even more rampant.

Google's Project Gemini ?aims to take generative AI to the next level with advances like factual grounding, reasoning skills, transparency (showing its work and explaining its reasoning, making it more transparent how it arrives at its answers), and versatility that current models lack. This could significantly improve the usefulness and trustworthiness of AI assistants. However, challenges like bias, misinformation, and social impacts will require ongoing research.

As AI systems grow more advanced at analysing and predicting human behaviour, safeguards and regulations will become critical. There are risks of centralised behavioural data being abused without user consent or transparency. Strict governance regarding consent, explainability, and rights over behavioural data access and usage is needed.

However, Regulation may need to catch up to AI capabilities. Ideally, regulatory frameworks needed to be proactive rather than reactive. The data used to train current models has already been processed without oversight. And AI could continue evolving behavioural analysis skills by ingesting real-time social data streams.

Data rights, auditing mechanisms, and independent ethical oversight should have been requirements for deploying AIs, not considered as an afterthought.?

Moreover, behavioural prediction AI could become a new arms race that countries and powerful companies will be reluctant to lose out on. Just as nations competed for nuclear capabilities, advanced AI behavioural models may be seen as an asset too strategically essential to regulate or restrict, regardless of the ethical concerns. More so than with past technologies, the ship may have already sailed in terms of controlling the advancement of AI in behavioural analysis and persuasion.

Accountability measures and ethical frameworks will also need to guide how behavioural insights can be applied. The public should be empowered to make informed choices about if and how their behavioural data is collected and leveraged. However, individual consent may not matter if these technologies are deemed crucial to the state and economic interests, as has happened historically with other disruptive innovations.

Summary

In summary, the rise of generative AI brings both promise and peril. While lowering barriers to information has benefits, we must be aware of the need to validate, verify, and think critically about the answers provided to avoid potential harm. Work will likely shift from finding solutions to validating and contextualising AI's responses. By embracing the positives while mitigating the risks, we can work to steer these technologies towards humane outcomes.

------

(1) Estimation based on the?total number of IT professionals worldwide ?and a range of wages per year between $20 - $100K.?

For example:?

Assuming the average wage is $100K annually, the total earnings would be 55.3 million * $100,000 = $5.53 trillion annually. 37% (i.e. the share of industry that could be automated) of this number would be $2.04 trillion annually.

It is important to note that these are just estimates, as the actual average wage and total earnings of the IT sector will vary depending on several factors.