 |
| Source: Teradata. Woods. |
"Everyone is tremendously excited! Boards ask CEOs, 'What is your AI strategy?' In turn, those officers are asking their teams to AI roadmap and get started. Like a shiny new toy, and like the folklore of 'there’s gold in them hills,' it has the lure and misconception of being able to solve nearly any problem that you can think of," said Jacqueline Woods, CMO, Teradata.
The authors of an IBM trend report for 2024 noted that 43% of CEOs said their enterprises are already using gen AI to inform strategic decisions, and 36% say the technology is being used for operational decisions. "In 2024, we expect to see continued increases," the authors of the report stated.
Forrester has seen similar enthusiasm. The research firm said that the number of enterprises in the experimentation and expansion stages of implementing gen AI jumped from 62% to 71% between July and September 2023, representing "one of the fastest mass adoption rates of a new technology in the enterprise".
"Fuelled by its potential to spur unprecedented innovation, the outlook for the 2024 Southeast Asian business landscape is characterised by enthusiasm for accessible and secure AI-infused decision-making technology. IDC predicts that by 2026, tech providers will allocate half of R&D, staffing, and investments to AI and automation. Our Alteryx-commissioned Enterprise of the Future survey echoes this sentiment.
"Seventy-eight percent of Singapore business leaders say that AI is already impacting what their organisations can achieve, and 41% state they will invest in these advanced technologies to respond to the changing market environment," said Philip Madgwick, Senior Director, Asia, Alteryx.
While the optimism is likely justified, the hype will die down in 2024, say industry observers, with the focus on applications that make an impact.
 |
| Source: Syniti. Ahlstrom. |
"Generative AI will plunge from the peak of inflated expectations to the trough of disillusionment. The consequence of the excitement around generative AI is that some companies have adopted the technology simply to keep up with the status quo rather than to solve a specific problem.
"As a result, we’re likely to see a great deal of investments in failed generative AI projects — hence, the fall into the trough of disillusionment," noted Rex Ahlstrom, Syniti's CTO.
"The key to limiting these failed projects is for companies to really ensure they understand the impetus and objectives for using generative AI; that it’s tied to a defined business outcome and there’s a method established for measuring the success of the investment."
“While gen AI has sparked incredibly creative ideas of how it will transform business and the world, there are very few real-world, scaled gen AI activities. As we move into 2024, we will see the first wave of gen AI enterprise projects reach levels of maturity that will expose important dimensions of gen AI not yet understood in the early phases,” said John Roese, Global CTO, Dell Technologies.
 |
| Source: Rackspace. Bhargava. |
Sandeep Bhargava, SVP, Global Services and Solutions, Public Cloud Business Unit at Rackspace Technology, predicted gen AI usage will evolve. "The focus on ethical usage, encompassing accountability, reliability, explainability, security, and privacy, will persistently grow, paving the way for policy implementations and solutions in these critical areas," he said.
Joe Ong, VP and GM, ASEAN, Hitachi Vantara, agreed, saying that conversations on AI in 2024 will revolve around "a mindful consideration of its challenges, as well as the need for trusted data and explainable AI".
 |
| Source: Digital Realty. Sharp. |
Chris Sharp, CTO, Digital Realty, advised enterprises to do their due diligence. "While there is a rush to adopt and scale AI workflows, it’s important that enterprises go slow to ensure their long-term success. They need to understand where AI technology is and how it's evolving before just diving in," he said.
Data quality
"Looking ahead, organisations will need to transcend the limitations of one-point solutions and harness the power of generative AI across an array of data sources. Since the technology is only as strong as the data it’s fed, this AI advancement demands the establishment of modernised data platforms, forming the foundation for seamless integration and application of generative AI," said Andrew Lim, MD, Kyndryl ASEAN.
Seagate also believes that more data will be a focus in 2024. "Businesses will be saving more operational data to teach AI, machine learning, and deep learning models moving forward; more companies will train models on both external and internal data so they can benefit from their proprietary information," the company stated in list of predictions for the year.
 |
| Source: AutoStore. Schitter. |
"The efficacy of AI systems inherently hinges upon the quality of the input data they receive. These systems depend on precise, comprehensive, and current information to render accurate forecasts and decisions. Flawed input data could compromise the AI system's outputs, leading to misdirected deliveries, improper resource allocation, or faulty customer communications.
"Such outcomes possess the potential to tarnish a company's reputation and undermine customer trust," warned Philipp Schitter, VP of Business Development APAC at AutoStore.
"The severe lack of harmonised and trusted data will continue to drive negative bias and poor customer experiences, resulting in underperforming business outcomes. To take full advantage of what AI has to offer, companies must undergo a radical data transformation to ensure their data is clean and pristine. Only then will companies be able to get to trusted data that leads to usable information and valuable insights to accelerate better business performance," agreed Woods.
"One crucial component that is often overlooked in the AI conversation is the quality of data. AI is not a magic bullet, and the results are only as good as the quality of data on which a model is trained," Sharp said.
"Ensuring that data is clean, accessible, and secure can be challenging and time-consuming. Data
scientists spend about 80% of their time collecting, cleaning and preparing data for analysis.
However, once an organisation has clean data, it will find it easier to maximise what it can do with it
and extract valuable insights," he said.
Ahlstrom from Syniti also predicted that increased adoption of generative AI will drive a need for clean data. "Data forms the foundation of generative AI, and for its potential to be harnessed, it needs clean data. Regardless of where you’re pulling the data from – whether you’re using something like modelling or a warehouse of your choice – quality data will be essential," said Ahlstrom.
"It’s the classic 'garbage in, garbage out' scenario. Bad data can result in bad recommendations, inaccuracies, bias, and so on. As more companies work to use generative AI, having a strong data governance strategy will become even more important. Ensuring your data stewards can access and control this data will also be key."
Data quality will start to become an executive-level topic in 2024, Ahlstrom added. "While ownership of data and data quality are core to business success, executive and board levels of most organisations often overlook this fact. We recently surveyed executives and found this played out in a discrepancy between perception and reality," he shared.
"Over 80% of executives surveyed believe they trust their data, but the reality is that people have to work hard to improve data quality to a level where the data can be consumed and used. As data quality takes on greater importance, it will escalate to an executive-level conversation."
Ahlstrom also said that the shift to data fabric will accelerate
due to
AI. He had anticipated that more organisations would move from a data
mesh approach to a data fabric a year ago as it would help flatten
information silos and make data available to business users more
quickly. "While the transition has been slow, this trend is certainly
accelerating. In 2024, this trajectory will be driven largely by
increased adoption of AI and other self-discovering technologies. Data
fabric has been a topic of conversation for several years, and now that
more advanced AI has emerged, it will become a bigger goal for
organisations," he said.
Gavin Barfield, VP & CTO, Solutions, Salesforce ASEAN, pointed out that many ASEAN organisations still have untapped data, driving more companies to lay the groundwork for AI by 2024. "A focus on data harmonisation, eliminating silos to extract value from data sources, democratising data for personalised insights, and a strong focus on data privacy and compliance are key priorities for corporate leaders," he said.
"This requires businesses to put in place strong data governance frameworks, well-defined policies for data collection, storage, and use, and constant monitoring and auditing of training data."
first mindset. Simply put, this is because getting the most out of tools like generative AI hinges on having
data in order - which entails good data management practices, including data access, hygiene and
governance," said Amir Sohrabi, Regional VP & Head of Digital Transformation, Emerging EMEA and Asia, SAS.
"Organisations that lack these best practices will find that generative AI isn’t as productive for them
because it does not have accurate and well-managed data to act on. Good data management also improves customer experience, and facilitates innovation, while positioning the business to capitalise on cost-saving opportunities. Expect CIOs to take the reins to empower organisation-wide data literacy."
Data infrastructure*
 |
| Source: Alteryx. Madgwick. |
"With data distributed globally across clouds and multi-tenant data centres in a hybrid IT infrastructure, flexible, on-demand, and open interconnectivity solutions are critical for extracting value from data. Additionally, data centres are much more likely to future-proof your deployments, from large-capacity runways for flexibility, to support for high-density AI deployments."
"As LLM usage becomes more pervasive across the business, it accentuates the imperative for data-centric infrastructures, necessitating substantial investments in robust data management systems. Forward-thinking IT leaders already see a direct correlation between facilitating value extraction from data at the speed and scale needed for real-time intelligence and AI-ready transformation," noted Madgwick.
“AI and sustainability are going to continue to influence technology adoption across Asia Pacific & Japan. Particularly for AI, there are potential headwinds such as the talent crunch but more importantly, without the right data infrastructure, AI projects might not yield the desired results and this could dampen enthusiasm,” agreed Matthew Oostveen, VP and CTO, Asia Pacific & Japan, Pure Storage.Data privacy
 |
| Source: Twilio. Adeniji. |
"Amid data breaches and data privacy concerns, we see organisations shifting towards data minimisation as a tactic to mitigate risks and adhere to privacy standards. However, there is a question on whether gathering only critical data for specific business objectives – instead of accumulating vast quantities of data – will impede the growth and effectiveness of AI-driven strategies. 2024 will see organisations exploring a feasible middle ground that ensures an optimal balance between data utility, privacy, and security."
Specialised AI models
 |
| Source: Hitachi Vantara. Ong. |
"For businesses to excel in employing generative AI, precise and reliable data is crucial, necessitating the establishment of a high-quality data foundation at the enterprise level. This underscores the importance of utilising private data in LLMs while also capitalising on publicly-available LLMs, all while ensuring data protection."
Barfield also predicted that the introduction of smaller and more specific LLMs will occur in 2024. "While foundational LLMs will be the backbone of generative AI, they are generic, designed for a mass audience, and expensive to run. With these constraints, more companies will start developing and using a combination of smaller, domain-specific language models for cost, performance, and latency reasons," he said.
"These specialised LLMs will be trained on specific datasets that can provide generative AI assistance in different and unique industries. For instance, a specialised model for the medical tech industry in Singapore will contain a wealth of data focused on medical knowledge and terminology, grounded in the local context.
"Across ASEAN, we may also see models fluent in local languages, such as those that can generate content in Bahasa Indonesia, Thai or Vietnamese. Specialised models will not need to store large amounts of data for general knowledge questions, and the smaller database will be more cost and energy-efficient to run for specialised industries."
 |
| Source: ManageEngine. Ganesan. |
Rajesh Ganesan, President at ManageEngine, said that the narrow applications of AI and its immense engineering difficulties call for AI training models that can cater to all aspects of a business. "Enterprise-focused LLMs help both employees and customers alike achieve deep-nested conversations with the enterprise's offerings and align better with evolving software tools.
"By adapting such models, enterprises will be better able to deploy their vast amount of knowledge to address both their creative and redundant workloads. It will also empower organisations to protect their data, reduce biases in their data, and provide detailed audit reports to understand AI decisions," he said.
Manuvir Das, NVIDIA's VP, Enterprise Computing, agreed that one size cannot fit all. In a blog post of 2024 predictions, he said: "Customisation is coming to enterprises. Companies won’t have one or two generative AI applications — many will have hundreds of customised applications using proprietary data that is suited to various parts of their business.
"Once running in production, these custom LLMs will feature RAG capabilities to connect data sources to generative AI models for more accurate, informed responses. Leading companies like Amdocs, Dropbox, Genentech, SAP, ServiceNow and Snowflake are already building new generative AI services built using RAG and LLMs."
*RAG stands for retrieval-augmented generation. The approach includes known third-party facts in AI-generated responses to improve accuracy.
Model optimisation
"How quickly and effectively an AI model can make predictions or decisions will be affected by
model optimisations, hardware and software optimisations including CPU accelerators, and data
preprocessing. To succeed at making improvements to AI models, I urge businesses to deploy
them at the edge. But work does not stop there, as these models will need to be monitored for
data and model drift once they are in action. The challenge for businesses would be to address
any changes in the data or model, and integrate these deployments into existing governance
models," said Jay Jenkins, CTO, Cloud Computing, Akamai.
"By ensuring continuous improvement, businesses will have the potential of surpassing previous
thresholds of success. Model supply chains, A/B testing, and feature store management
frameworks will also need to be developed or integrated. For organisations using multicloud
services and dealing with various data sources, they cannot afford to ignore flexible
architectures to avoid cloud concentration risks and lock-in."
Consent for training data
Ong noted that gen AI usage comes with risks. "While there's a growing inclination in certain scenarios to maintain both applications and data on-premises rather than opting for cloud-based solutions, it's essential to acknowledge the associated risks," he said.
"Notably, when information shared with public LLMs becomes accessible, concerns about external parties potentially benefiting from original data or code may arise. Going into the year 2024, this may prompt calls for regulations to avoid issues of plagiarism and copyright."
Accountability |
| Source: Salesforce. Barfield. |
"Who is responsible for the advice that a chatbot makes to the customer? And what happens when a chatbot gives misguided recommendations to the customer? There remain considerable risks of a chatbot giving inappropriate responses that could result in serious liability issues for the company. Human supervision will still be needed as a check and balance to validate and correct any inaccurate results generated by AI."
The problem of maintenance
Mona Ghadiri, Senior Director of Product Management, BlueVoyant, touched on how AI solutions are designed. She said: "It’s interesting to think about how we build teams to design AI. AI demands operationalisation. So, you built it, how are you going to maintain it? In 2024, we will see that the designer is not going to be the caregiver. There will need to be another DevOps-esque intersection of those making sure AI is running properly.”
Python will also be on developers' minds when it comes to AI, said Laurent Doguin, Director of Developer Relations and Strategy at Couchbase. "Improving developer productivity is critical to the success of any technology organisation. Each developer has different tasks, different stacks to work with and requires different things from various LLMs. As a result, developers will need to start fine-tuning LLMs to their specific needs, and Python is the default programming language to achieve that," he said.
Scaling AI
"When it comes to a successful AI strategy, it’s not practical to have a cloud-only approach. Companies need to consider an edge computing strategy – in tandem with the cloud – to enable low-latency, real-time AI predictions in a cost effective way without compromising on data privacy and sovereignty," said Priya Rajagopal, Director of Product Management at Couchbase.
 |
| Source: Lenovo. Mitra. |
"With the proliferation of data, organisations will need to have efficient edge computing infrastructure to overcome AI deployment complexities and quickly translate data into actionable insights that streamline operations and improve business outcomes," agreed Kumar Mitra, MD and Regional GM, Central Asia Pacific at Lenovo ISG.
"I predict that the convergence of edge and cloud AI is the way to deliver AI at scale with the cloud and edge offloading computational tasks to the other side as needed. For instance, the edge can handle model inferences while the cloud may handle model training or the edge may offload queries to the cloud depending on the length of a prompt and so on," Rajagopal added.
She also noted that edge AI can only be viable option if AI models are "lightweight and capable of running in resource-constrained embedded devices and edge servers while continuing to deliver results at acceptable levels of accuracy".
"Models need to strike the right balance — meaning, models must be small and less computationally intensive so they can run efficiently at the edge while also delivering accurate results. While a lot of progress has been made in model compression, I predict that there will be continued innovation in this space, which when coupled with advancements in edge AI processors will make edge AI ubiquitous," she said.
Responsibilities
There are other hard questions that need to be asked, Ghadiri said. “As AI has become more profitable there has been a shift of mindset using AI for financial goals and ethical use cases... I think as we consider the concept of profitability, cyber becomes this big question mark. Whose job is it to make this secure, yet also ethical?" she asked.
Mitra of Lenovo listed various ethical dilemmas brought about by AI, such as lack of transparency, gender and ethnic bias, as well as threats to privacy, danger of mass surveillance, and the growing use of unreliable AI technologies in law enforcement. "This makes it imperative to have a global ethical AI framework," he said.
Legislation
 |
| Source: NUS. Pang. |
"With the growing popularity of AI technology in content creation, autonomous driving, medical diagnosis, and other fields, legal and compliance issues are becoming increasingly important," he said.
"Regions such as North America, Europe, and Asian nations like China, Japan, South Korea, and Singapore are rapidly developing and refining AI regulation frameworks. However, for many developing countries, AI regulation is still in its nascent stage, with the main challenge being how to strike a balance between technological innovation, ethics and safety," he observed.
"Some countries have begun to formulate basic AI strategies and policies, but most have yet to establish a comprehensive regulatory framework. In addition to domestic regulatory measures, international organisations like the United Nations, the World Economic Forum, and the Organisation for Economic Co-operation and Development (OECD) also need to actively promote global cooperation in AI regulation. These organisations primarily focus on the impact of AI on the global economy, society, and security, and are committed to building a fair, transparent, and sustainable global AI ecosystem.
"Looking ahead, regulations governing artificial intelligence worldwide will continue to evolve and improve to adapt to the rapid development and global application of AI technology. At the same time, the regulatory process needs to balance multiple aspects such as technological innovation, privacy protection, ethics, and social responsibility to ensure the sustainable development and application of AI technology."
"While Singapore is not looking to urgently regulate AI, in (2024) we expect increased government
involvement that CISOs must prepare for, as countries promote more responsible use of AI – which could include continued AI regulation, and new post-quantum guidance," predicted James Cook, Director of Digital Security, Asia Pacific & Japan at Entrust.
"Although the nation is taking a more wait-and-see approach, businesses should consider each of these a call to action to improve not only their own cybersecurity strategies, but also to consider the impact of new technologies, like AI, on their organisation and their customers.
"An uptick in government guidance will help create a blueprint for businesses to navigate rising challenges and security threats. But understanding and complying with the anticipated patchwork of regulations and regional legislation may pose a challenge for businesses, especially those operating across borders. CISOs and leaders will need trusted advisors, sound support, and secure solutions to successfully and safely forge ahead.”
 |
| Source: UiPath. O'Reilly. |
"For instance, Singapore’s Infocomm Media Development Authority (IMDA) and the AI Verify Foundation recently launched the Gen AI Evaluation Sandbox to support new benchmarks for evaluating generative AI," she said.
"Amid evolving data privacy and protection regulations around AI, the C-suite will take serious steps to counter the potential risks for AI misuse and miscalculation. Consequently, this will give rise to new safeguards and innovations that will refine the AI risk-benefit equation."
O'Reilly also noted that effective AI governance will become paramount when it comes to achieving robust AI outcomes. "In 2024, an increasing number of organisations will witness the evolution of AI governance from aspiration to implementation guided by innovation as enterprise software companies build AI controls into their own offerings," she said.
"AI providers and scientists will shift their focus towards constructing additional layers of trust, so organisations can confidently leverage new AI capabilities with the knowledge that their data is secure."
Getting involved
Many enterprises are still hesitant about gen AI, according to the IBM trends report, noting that more than 60% of enterprises have not yet developed a consistent, enterprise-wide approach to gen AI. Ghadiri also observed that no one can become an expert on AI in business without actually trialling AI. "The only experts in AI thus far are those adapting to the change," she pointed out.
Her advice for those yet to embrace AI: "While many will say it’s overkill because we see it in the headlines every day, the truth is we just need to get past the initial hype phase. Right now it’s easy to be self-conscious, but once we get to the 2nd, 3rd and 4th generations of this thing, we can comfortably benefit from it because we fully understand its risks.”
 |
| Source: Kissflow. Varadharajan. |
Roese summed it up: "The centre of the universe for 2024 is AI. (2024) will be all about putting AI into practice at the edge, diversifying the hardware pool and securing it through Zero Trust. We’re entering an AI era so need to recognise that long term, AI will drive architectural change across the entire IT ecosystem for many years to come.
"Actively think of AI but do not do it independent of other architectures – this is how you’ll make sure your visions and actions align for long-term success.”
Tread carefully
"For now, generative AI and large language models have given us a glimpse of how businesses, industries, and society can be transformed for the better. But with all the excitement comes FOMO – fear of missing out – and the companies that rush to implement AI initiatives without patience and careful planning can risk damaging their corporate reputation," Woods concluded.
"Rather than jumping on the bandwagon, you need to be the driver’s seat of your AI journey. Companies that do so carefully and authentically will be the ones that come out ahead."
ASEAN opportunity
 |
| Source: Salesforce. Abraham. |
"Businesses will need access to trusted technology, AI training and guard rails for responsible AI adoption," summarised Sujith Abraham, Senior VP and GM, ASEAN for Salesforce.
"Done right, businesses can harness the full potential of AI supercharge growth in the region."
Abraham also said that there is a significant opportunity for ASEAN businesses to leapfrog and elevate themselves to the global stage with AI. "ASEAN countries have made steady progress in preparation for AI adoption – Singapore currently leads the APAC region in overall AI readiness, while Thailand and Indonesia recorded one of the largest improvements in government AI readiness.
"New AI policy initiatives such as the ASEAN Guide on AI Governance and Ethics, and Singapore’s IMDA AI Verify Foundation, will further serve as a torch for countries to deploy responsible and innovative AI," he concluded.
Barfield said ASEAN organisations will start small on AI. "Despite all the progress, businesses in the region are cautiously exploring generative AI. Besides trust and accountability concerns, many are still still building their data foundation. In the near future, we may only see select industries test the waters with internal deployments before venturing out with smaller-scale customer service programmes," he said.
*More insights on data centres and AI can be found in D is for data and data centres at https://www.techtradeasia.com/2024/01/a-to-z-of-tech-predictions-in-2024-d-f.html
No comments:
Post a Comment