Hi Everyone,

We always like you to get the most value out of the newsletter, in addition to the email version of the newsletter there is a supporting website where you can log on and get to all the previous versions.

A slightly different format this week. A number of in depth reports have been released this month, discussing the current ‘state’ of AI adoption and expected impacts. We are going to just have one deep dive on this information.

The state of AI in April 2024

Amongst the various impact reports, and perhaps a good place to start is the summary of the Google NEXT conference. It was absolutely dominated by AI product and examples. Reception of the various announcements though have been muted. It isn’t that a large amount of ‘good’ technology was not released, however a lot is ‘in preview’, ‘for release in the summer’. It still feels as though Google are trying to catch up with the market leaders, rather than asserting a leading position. It will reflect the focus on big technology continuing to lead with AI driven product announcements, and opportunities to increase the enterprise license costs. For example Googles inbuilt ‘otter transcribe’ type feature, licensed on an additional cost per seat, as opposed to a new feature in your existing stack.

The key CIO questions at the moment are

• how do we simplify the technology stack (and license costs)

• how do we stop some activity

• how do we get a semantic and comprehensive view of the total enterprise available data, and unlock it

These are all the underpins to making space for AI to fit into the already full transformation activities in most large enterprises.

In small to mid tier organisations, the current and most common questions are still ‘where and how to get started’, ‘what do I need to buy, if anything’ and ‘how do we move from interesting chatBots about recipes to enterprise productivity gains’. Probably as reflected by the following deep dive, the question everyone should be asking is ‘how do we reskill the workforce and how do we reinvent the processes in the business using these new skills and tools’.

Let’s dive into the two key reports.

Transformed by AI

The report primarily discusses the implications of generative AI on the workforce, focusing on both the potential enhancements and disruptions that AI technologies might bring to various jobs. It elaborates on the concept of AI exposure—how certain tasks and professions are more likely to be affected by AI advancements.

The study reveals that 11% of tasks in the UK, when scaled by hours worked, are already exposed to 'here and now' generative AI systems that are widely available. Looking ahead, the exposure jumps to a staggering 59% of hours worked in the UK when considering more advanced 'integrated' AI systems that can access proprietary data and execute tasks autonomously.

This level of projected impact is on par with the transformation that digitalization has brought to knowledge work since the 1990s. The IPPR's findings align with a recent IMF study that found a similar scale of exposure.

While these shifts will not happen automatically, the immense capabilities and commercial benefits of generative AI will likely drive strong adoption. A Goldman Sachs survey found that while only 5% of CEOs expect AI to significantly impact their business in the next 1-2 years, a full 65% believe it will have a major impact within 3-5 years.

This is considered over four phases (starting at phase zero, which is now)

2024 marks the move into phase 1.

Phase 1 focuses on 'low-hanging fruit' use cases, where tasks can be completed more quickly using generative AI without significant process changes. This might include using AI to extract information from already digitized databases or texts.

Phase 2 involves integrating AI systems into organizational processes. While this can be done through generative AI, it requires some re-engineering of workflows. Examples include AI ordering supplies for a restaurant kitchen (requiring employees to log stocks in an app) or assisting teachers with grading (necessitating the digital submission of student work).

Phase 3 entails rebuilding processes around AI, including tasks that would require changes to social norms or significant regulatory shifts. This could encompass AI systems advising patients on treatment plans or making personnel decisions.

Crucially, the report highlights that society can actively shape the impact of generative AI on the labour market through design choices. Certain tasks in teaching, policing, therapy, or medical advice could be 'ring fenced' to ensure a 'human touch', even if they could theoretically be automated. Maintaining higher employment levels in this way could lead to greater economic output compared to maximizing automation.

The IPPR report presents several potential economic impacts of generative AI, which can be framed in terms of opportunities and risks:

Opportunities:

Productivity gains: In the report's central scenario, the adoption of generative AI could lead to significant productivity increases in certain occupations. For instance, secretarial and related occupations could see productivity gains of 35%, while customer service occupations could experience a 32% boost.

Economic growth: The report estimates that if generative AI is widely integrated across the economy, it could provide an economic boost of up to 13% of GDP in the 'full augmentation' scenario, where AI enhances worker productivity without job displacement.

Creation of new jobs and tasks: As seen in previous technological revolutions, the adoption of generative AI could lead to the emergence of entirely new jobs and tasks. The report suggests that a well-designed industrial strategy could help create demand for jobs that are less exposed to automation.

Risks:

Job displacement: In the 'full displacement' scenario, where generative AI replaces workers without creating new jobs, the report estimates that up to 8 million jobs could be lost in the UK, with no corresponding GDP gains.

Unequal impact across occupations: The study finds that certain occupations, such as back-office jobs and those predominantly held by women, are more exposed to potential job displacement in the initial phases of AI adoption.

Widening inequality: If the benefits of AI-driven productivity gains accrue primarily to business owners and a smaller number of high-skilled workers, it could exacerbate income and wealth inequality.

Transitional challenges: As generative AI disrupts the labour market, workers may face difficulties transitioning to new roles, particularly if their skills do not align with the demands of emerging jobs. This could lead to increased unemployment and social dislocation in the short to medium term.

In conclusion

Based on the IPPR report, the most likely central outcome of generative AI's impact on the UK labour market is a mixed scenario where some jobs are displaced while others are augmented, leading to moderate job losses but significant productivity gains. In this scenario, approximately 4.4 million jobs could be lost, but the economy could still see a GDP boost of around 6.4% due to increased productivity.

The report identifies certain groups as being most at risk of job displacement, particularly in the initial phases of AI adoption. These include workers in back-office roles, entry-level positions, and part-time jobs. Women are also disproportionately exposed, as they are more likely to hold jobs in highly impacted occupations such as secretarial and administrative roles.

To ensure the most positive outcome, business leaders should focus on a multi-faceted approach that prioritizes reskilling, job augmentation, and supporting workers through transitions. Key steps include:

1. Investing in reskilling and upskilling programs: Businesses should work with educational institutions and government agencies to develop targeted training initiatives that equip workers with the skills needed to thrive in an AI-driven economy. This will help resolve the reskilling risk and facilitate smoother job transitions.

2. Focusing on job augmentation rather than replacement: By designing AI implementation strategies that prioritize enhancing human capabilities rather than replacing workers outright, businesses can harness the productivity benefits of generative AI while minimizing job losses.

3. Providing support for job transitions: Employers should offer guidance, resources, and assistance to help workers navigate the changing job landscape and transition to new roles that are less susceptible to automation.

4. Ensuring equitable distribution of AI's benefits: Business leaders should work to create an environment where the gains from AI-driven productivity increases are shared fairly among workers, rather than concentrating solely in the hands of owners and top executives.

5. Collaborating with policymakers and other stakeholders: To create a comprehensive, society-wide response to the challenges posed by generative AI, business leaders should engage with government officials, labour organizations, and civil society groups to develop policies and initiatives that promote inclusive growth and mitigate potential negative impacts.

By proactively addressing the risks associated with generative AI and focusing on strategies that prioritise worker well-being and adaptability, business leaders can help steer the UK economy towards a future where the benefits of this transformative technology are widely shared and its disruptive potential is effectively managed.

The AI Index Report

This report has a more ‘technical’ rather than economic impact view, and is also based around USA data, rather than UK. It still provides for some interesting, and sometimes startling information. (NB its 500 pages, so have a good supply of coffee available).

The top takeaways

1. AI beats humans on some tasks, but not all. Versus several benchmarks, visual reasoning and image classification, but still trails in complex reasoning

2. US, leads china and Uk as leading sources of AI

3. Standardisation of AI responsibility are lacking

4. Generative AI investment continues to grow, reaching $25Billion

5. AI Makes workers more productive and leads to higher quality work

6. AI is helping with Scientific Progress

7. People are more globally aware of AI

AI research continues to increase in rate, this is reinforcing that we are still at an early stage of AI. although last year the increase was 1.1% year on year. Compared to other areas the amount of research is prolific.

The complexity and size of LLM AI models has continued to grow, leading to some concerns that the amount of available data ‘might run out’. The trend for this continues, although promising research around the power of smaller mixed mode or specialist models is promising. These large foundation models are the backbone of large Tech AI providers.

As of 2023, there are still some task categories where AI fails to exceed human ability. These tend to be more complex cognitive tasks, such as visual commonsense reasoning and advanced-level mathematical problem-solving (competition-level math problems), complex task reasoning.

Over the course of 2023, the growth of multimodal (video, audio, text) has continued to provide new tools and new ways of solving problems. 2024 is likely to continue this trend with new tools creating music, complex video, and animations already announced.

One of the more interesting benchmarks released is called “MMMU: A Massive Multi-discipline Multimodal Understanding and Reasoning Benchmark for Expert AGI“,

This is one of the current few benchmarks which AI is still far behind human capabilities.

In addition new benchmarks are now also available for measuring the performance of AI Agents, which as this is the likely next wave of process automation technology, will be useful to consider.

Impact on employment

The number of jobs requiring AI skills is around 1.5% (slightly down versus last year), although that needs to be considered in line that overall job postings are down, including the downturn on technology roles, still correcting from the COVID over staffing in large technology companies.

The latest McKinsey report reveals that in 2023, 55% of organizations surveyed have implemented AI in at least one business unit or function, marking a slight increase from 50% in 2022 and a significant jump from 20% in 2017 (Figure 4.4.1). AI adoption has spiked over the past five years, and in the future, McKinsey expects to see even greater changes happening at higher frequencies, given the rate of both AI technical advancement and adoption

In the past year, several studies have provided compelling evidence regarding the impact of AI on productivity in the workplace. The findings suggest that AI tools can significantly enhance worker efficiency and output quality.

For instance, Microsoft conducted a comprehensive analysis by aggregating multiple studies that compared the performance of employees using AI-powered tools like Microsoft Copilot or GitHub's Copilot with those who did not have access to such tools. The meta-review revealed that workers utilizing these LLM-based productivity tools completed their tasks between 26% and 73% faster than their peers who did not use AI assistance.

Impact on workforce size in the next 3 years:

• 30% of respondents expected little to no change in the number of employees

• 43% believed that staff size would decrease

• Only 15% anticipated an increase in the number of employees

Expectations for employee reskilling:

• 38% of respondents indicated that 6-10% of their workforce would require reskilling

• 18% believed that 11-20% of their employees would need reskilling

• 17% expected that more than 20% of their employees would require reskilling

Its anticipated a decrease in staff size due to AI than those who expect an increase, A significant portion of the workforce is expected to require reskilling to work effectively with AI technologies.

In conclusion

As a business leader, the AI Index 2024 report highlights several key takeaways and strategic considerations:

1. Invest in AI adoption: With AI surpassing human performance in various tasks and boosting worker productivity, businesses should strategically invest in AI technologies to remain competitive and improve efficiency.

2. Monitor industry trends: As industry players dominate frontier AI research, business leaders must stay informed about the latest advancements and trends to make informed decisions about AI adoption and partnerships.

3. Allocate resources wisely: The increasing costs of training state-of-the-art AI models emphasise the need for businesses to allocate resources wisely and consider collaborations to access cutting-edge AI technologies.

4. Prioritise responsible AI: The lack of standardised evaluations for assessing the responsibility of AI systems underscores the importance of prioritising responsible AI practices within organisations to mitigate potential risks and maintain public trust.

5. Capitalise on generative AI: With the surge in funding for generative AI, businesses should explore opportunities to leverage these technologies for innovation and growth.

6. Foster human-AI collaboration: Studies demonstrating AI's potential to enhance worker productivity and bridge skill gaps highlight the need for businesses to foster effective human-AI collaboration and provide adequate training for employees.

7. Navigate regulatory landscape: As AI-related regulations increase, business leaders must navigate the evolving regulatory landscape and ensure compliance while advocating for policies that support innovation.

8. Address public concerns: With growing public awareness and nervousness regarding AI's impact, businesses should prioritise transparency, accountability, and effective communication to address public concerns and maintain trust.

In summary, business leaders must proactively adapt their strategies to leverage the potential of AI while navigating the challenges and risks associated with its adoption. This requires a combination of strategic investment, responsible AI practices, effective human-AI collaboration, and active engagement with stakeholders, including policymakers and the public, to ensure that AI is developed and deployed in a manner that benefits both businesses and society as a whole.

Considering the common themes of Both reports.

Based on both the IPPR and Stanford University reports, there are three key shared conclusions for business leaders to act upon:

1. Embrace AI and reskilling: Invest in AI technologies and prioritise upskilling your workforce. Provide training and support to help employees transition, harnessing AI's benefits while minimising job displacement.

2. Foster human-AI collaboration: Design AI strategies that augment human capabilities rather than replace workers. Create an environment where humans and AI work together effectively, leading to higher quality work and better outcomes.

3. Prioritise responsible AI: Ensure transparency, accountability, and fairness in AI development and deployment. Engage with policymakers, stakeholders, and the public to address concerns, maintain trust, and contribute to regulations that support innovation while protecting society.

4. The impact is going to be profound and disruptive.

Detailed Reading

The full report links here for completeness.