Produced in partnership with Many enterprises aiming to

disrupt their industries with

generative AI underestimate
the requirements for effective
deployment of the technology.

Generative AI:
Differentiating disruptors
from the disrupted
2  MIT Technology Review Insights

Preface
“Generative AI: Differentiating disruptors from the disrupted” is an MIT Technology Review
Insights report sponsored by Telstra International. To produce this report, MIT Technology
Review Insights conducted a global poll of 300 executives. The report also draws upon
eight in-depth interviews with data and AI experts, conducted in November and December
2023. Paul Kielstra was the author of the report, KweeChuan Yeo was the editor, and Nicola
Crepaldi was the publisher. The research is editorially independent, and the views expressed
are those of MIT Technology Review Insights.

We would like to thank the following individuals for providing their time and insights:

Geraldine Kor, Managing Director of South Asia and Head of Global Enterprise, Telstra
International

Laurence Liew, Director of AI Innovation, AI Singapore

Giuseppe Nuti, Head of Machine Learning and Artificial Intelligence, UBS Global Markets

Michael Schrage, Fellow, Center for Digital Business, MIT Sloan School

Stela Solar, Inaugural Director, Australia’s National Artificial Intelligence Centre,

Commonwealth Scientific and Industrial Research Organisation

Dayle Stevens, Executive – Data & AI, Telstra

Calvin Wong, CEO and Center Director, Laboratory for Artificial Intelligence in Design,
and Cheng Yik Hung Professor in Fashion, School of Fashion and Textiles, The Hong Kong
Polytechnic University
 MIT Technology Review Insights 3

CONTENTS

01 Executive summary.........................................................................4

02 Introduction: Great expectations..........................................7

Unusual calm about disruption....................................................8

03 Deployment in 2023:
Learning in pool’s shallow end...............................................10

04 2024: Year of grand ambition.................................................13

05 Ambition and hubris signal weak

readiness for rapid adoption...................................................15
IT resources and capabilities often fall short ..................15
Partner perspective..........................................................................17
Non-technological barriers pose daunting,
long-term challenges..................................................................... 20
Early adopters see things differently...................................22

06 Conclusion: Grand plans or unrealistic dreams......24

4  MIT Technology Review Insights

01 Executive
summary

G
enerative AI1, though still an emergent The overarching message from this research is that
technology, has been in the headlines since plans among corporate leaders to disrupt competition
OpenAI’s ChatGPT sparked a global frenzy using the new technology—rather than being
in 2023. The technology has rapidly disrupted–—may founder on a host of challenges that
advanced far beyond its early, human-like many executives appear to underestimate.
capacity to enhance chat functions. It shows extensive
promise across a range of use cases, including content Executives expect generative AI to disrupt industries
creation, translation, image processing, and code writing. across economies. Overall, six out of 10 respondents
Generative AI has the potential not only to reshape key agree that “generative AI technology will substantially
business operations, but also to shift the competitive disrupt our industry over the next five years.”
landscape across most industries. Respondents that foresee disruption exceed those
that do not across every industry. Rather than being
The technology has already started to affect various confined to specific sectors, the technology is likely to
business functions, such as product innovation, supply transform specific functions across the economy, such
chain logistics, and sales and customer experience. as corporate IT and customer service.
Companies are also beginning to see positive return on
investment (ROI) from deployment of generative
AI-powered platforms and tools.
“Generative AI technology will
While any assessment of the technology’s likely
substantially disrupt our industry
business impact remains more forecast than empirical, over the next five years.”

62
it is necessary to look beyond the inevitable hype.
To examine enterprises’ technological and business

%
needs for effective implementation of generative
AI, 300 senior executives across a range of regions
and industries were surveyed. Respondents were
asked about the extent of their corporate rollouts,
implementation plans, and the barriers to deployment.
Combined with insights from an expert interview panel, of poll respondents agree
this global survey sheds light on how companies may or
may not be ready to tackle the challenges to effective Source: Compiled by MIT Technology Review Insights survey, 2024

adoption of generative AI.

 MIT Technology Review Insights 5

the technology widely. Those that used the technology

experimented with or deployed it in only one or a
Poll methodology few limited areas. The most common use case was
automating low-value tasks—a use with low-to-
In November and December 2023, MIT Technology
modest gains but minimal risks. Expert interviewees
Review Insights polled business leaders about
characterize generative AI deployments as tactical
how organizations are implementing—or planning
experimentations rather than holistic transformations.
to implement—generative AI technologies, along
with the barriers to effective deployment of the
Companies have ambitious plans to increase
technology. The 300 respondents are C-suite
adoption in 2024. Respondents expect the number
executives (70%) and vice presidents or directors
of functions where they aim to deploy generative
(30%). The majority (70%) of respondents
AI to more than double in 2024. This will involve
manage information technology, data, and data
frequent application of the technology in customer
engineering-related functions. They are distributed
experience, strategic analysis, and product innovation.
among industries including financial services,
Respondents plan to increase use of generative AI
banking, and insurance (17%); consumer packaged
in fields relevant to their industries, such as coding
goods and retail (17%); manufacturing and
in IT, supply change management in logistics,
automotive (17%); technology and telecom (17%);
and compliance in financial services. Experts we
logistics (13%); energy, oil, and gas (10%); and
interviewed say companies will search for innovative
media and communications (10%). The responses
use cases that may give them unique competitive
are global in scope, representing the Americas
advantages, along with ongoing expansive rollout of
(17%), Europe (17%), and Asia-Pacific (66%).
generative AI.

Companies need to address IT deficiencies, or risk

A majority of respondents do not envision AI falling short of their ambitions to deploy generative
disruption as a risk; instead, they hope to be AI, leaving them open to disruption. Fewer than
disruptors. Rather than being concerned about risk, 30% of respondents rank each of eight IT attributes
78% see generative AI as a competitive opportunity. at their companies as conducive to rapid adoption
Just 8% regard it as a threat. Most respondents hope of generative AI. Moreover, these results may be
to be disruptors: 65% say their businesses are “actively overly optimistic. Those with the most experience of
considering new and innovative ways to use generative deploying generative AI—called early adopters in this
AI to unlock hidden opportunities from our data.” report—have less confidence in their IT than their
peers. At best, many early adopters (65%) say their
Despite expectations of change, few companies hardware is modestly conducive to rapid adoption.
went beyond experimentation with, or limited Approximately half say the same of data volume
adoption of, generative AI in 2023. Although (54%), accuracy (50%), and storage infrastructure
most (76%) companies surveyed had worked with (50%). These assets, however, are necessary to
generative AI in some way in 2023, few (9%) adopted develop and run the AI from which companies

“Effectively deploying generative AI solutions is

predicated upon having 100% confidence in the
end-to-end operationalization of capturing,
processing, contextualizing, and actioning data.”
Geraldine Kor, Managing Director of South Asia and Head of Global Enterprise, Telstra
International
6  MIT Technology Review Insights

0
seek to benefit. According to Geraldine Kor, Telstra • Competitive environment: Early adopters are more
International’s managing director of south Asia and than twice as likely to see the competitive environment
head of global enterprise, “effectively deploying as an enabler of rapid generative AI adoption than as a
generative AI solutions is predicated upon having barrier; among other respondents, the opposite is true.
100% confidence in the end-to-end operationalization Generative AI for its own sake makes limited business
of capturing, processing, contextualizing, and sense. Any adoption strategy needs to be consistent
actioning data—particularly if attempting to do so with a search for improved competitive standing.
in real time or near real time.” Such data, in turn, is
essential for getting full value from generative AI. • Culture: Early adopters are more likely to regard
Many companies need to rethink their underlying IT attitudes toward innovation at their companies as an
by, at least, creating an appropriate data strategy and enabler of rapid AI rollout; other surveyed executives are
securing reliable access to the necessary hardware more likely to report that this aspect of culture is a
for deploying generative AI. barrier. As with any change management program,
adoption plans need to include efforts to win over users
Non-IT factors also undermine the successful use of and executives.
generative AI. Survey respondents also report non-IT
impediments to the extensive use of generative AI. • Skills: The IT skills needed for significant AI projects
are in short supply but among our respondents, early
• Risk: 77% of respondents cite their regulatory, adopters are more acutely aware of the shortage of
compliance, and data privacy environment as a leading available talent. Wider business skills are also necessary
barrier to rapid AI adoption. Meanwhile, 54% of early for applying the technology well. Users of even basic
adopters admit their companies’ cybersecurity generative AI-powered tools need to understand their
measures are at best modestly capable of supporting a risks, limitations, and potential benefits. Talent
rollout. Any strategy to use the technology needs to be acquisitions and talent development are integral to
rooted within holistic governance structures that successful adoption.
address this inter-related complex of risk.
Executives expect generative AI to provoke a wave
• Budgets: 56% list IT investment budgets as a leading of disruption. In many cases, however, their hopes to
barrier. One expert advises executives to examine use be on the right side of this innovation are endangered
cases in terms of ROI, not just cost, to judge better what by impediments that their companies do not fully
spending is worthwhile. appreciate.

Non-IT factors that undermine deployment of generative AI

Risk Budgets Competitive Culture Skills

environment
Source: Compiled by MIT Technology Review Insights survey, 2024
 MIT Technology Review Insights 7

02 Introduction:
Great expectations

G
enerative AI dominated news headlines in “Like cloud, mobile devices, and the original world wide
2023, fueling dramatic forecasts, such as web protocols, generative AI is both foundational and a
the extent of disruption to businesses in force multiplier,” says Michael Schrage, research fellow
the near term and the significant market of MIT Sloan School’s Center for Digital Business.
opportunities that may soon be available. “There’s no escaping it and we’re not even close to the
McKinsey estimated in 2023 that adoption of generative end of its beginning.” Giuseppe Nuti, head of machine
AI across industries will drive $3.5 trillion in global learning and AI at UBS Global Markets, adds that
economic activity. That number could double if the generative AI “clearly has the potential to be highly
technology is used more widely.2 Meanwhile, Bloomberg impactful.” The technology’s ability to create content,
Intelligence projected in 2023 that the worldwide market address complex problems, automate, predict, and
for generative AI products will grow from $67 billion in interact easily with humans are both novel and useful.
2023 to $1.3 trillion by 2032, a compound annual growth
rate of 42%.3 As with other nascent technologies, hype can obscure
substance. Gartner puts generative AI at the top of its
Investors and senior executives are also increasingly 2023 Hype Cycle for Emerging Technologies, which
talking about generative AI. According to S&P Global precedes “the trough of disillusionment” phase.6
Market Intelligence, for example, during earnings calls This does not mean the talk is meaningless. In 2015,
of U.S. S&P 500 companies, mentions of generative AI for example, IoT was at the peak of Gartner’s 2015
rose tenfold between 2022 and 2023.4 Nearly half of Emerging Technologies Hype Cycle,7 and it is now
S&P companies mentioned the technology on these commonplace in business.
calls between May and December 2023—a proportion
similar to those who talked about interest rates during It’s risky to sit on the sidelines as others undertake the
calls, according to NBC News.5 transition from hype to value, says an AI expert at a

No sector doubts the likely impact of generative AI:

respondents from every industry agree generative AI
will bring disruption.
8  MIT Technology Review Insights

Figure 1: Generative AI is seen as a disruptive opportunity

90%
To what extent do you agree with the following statements? Agree Disagree
80%

70%
78%

60%
62%
50%

40%

30%

20% 26%

10%
8%
0%
Generative AI technology will substantially We see generative AI as more of a competitive
disrupt our industry over the next five years. opportunity than a threat.

Source: MIT Technology Review Insights survey, 2024

global IT consultancy. He likens the current situation to Predictably, executives from certain industries are
that of a peloton, the main pack of riders in a cycling more likely to anticipate change. For example, 87% of
race. Most companies, he says, are currently huddled respondents in media and communications—heavy
in the pack. “For innovators, their skills will get better, users of IT and producers of substantial creative
as their data becomes more and more available in content—say generative AI will bring disruption,
better quality,” he says. “As they are able to start compared with 7% who disagree. No sector doubts the
identifying more and more use cases, you will start likely impact of generative AI: respondents from every
seeing leaders emerge and pull away from the pack.” industry agree generative AI will bring disruption.
The risk for others is that without learning lessons by
staying in the race, they will lag further behind those “Generative AI is going to increasingly gobble up
working with the technology. any business involved in giving advice: technical,
spiritual, business, and financial,” says Schrage.
Unusual calm about disruption Stela Solar, inaugural director of Australia’s National
Many executives surveyed believe generative AI Artificial Intelligence Centre, goes further. She argues
will quickly and markedly shake up the business that the effect is unlikely to be highly concentrated
environment (see Figure 1). A majority (62%) agree that in a few industries. Instead, functions common to
“generative AI technology will substantially disrupt our many—such as customer service or finance—will
industry over the next five years.” Just 26% say the use the technology to generate value and enhance
opposite. productivity.

“Generative AI is going to increasingly gobble up any

business involved in giving advice.”
Michael Schrage, Fellow, Center for Digital Business, MIT Sloan School
 MIT Technology Review Insights 9

Areas where generative AI will have the biggest impact

Sales Marketing Customer Software Software R&D

operations development (product innovation)
(corporate IT)

Source: Compiled by MIT Technology Review Insights from McKinsey & Company, 2024

McKinsey expects generative AI to have its biggest While business leaders are usually wary of disruption,
effect in sales, marketing, customer operations, respondents appear more focused on possibilities than
software development (for corporate IT), software risks. Overall, most respondents (78%) say generative
(for product innovation), and R&D. Everywhere except AI is more of a competitive opportunity than a threat;
the public sector, 2023 McKinsey research expects 8% say the opposite. Other survey data suggests this
generative AI to highly impact at least one—sometimes response reflects respondents’ hopes to be on the side
several—of these fields.8 Meanwhile, academic of the disruptors. In particular, 65% say their business
research foresees the technology will impact common “is actively considering new and innovative ways to use
tasks in many sectors. A 2023 study of GitHub Copilot generative AI to unlock hidden opportunities from our
says generative AI-powered code development tools data, which would not have been possible with earlier
could increase web design speed by more than 50%.9 technologies.” Just 13% are not doing so.
A 2023 National Bureau of Economic Research
working paper predicts generative AI customer service A closer look at adoption of generative AI across
software could raise productivity by 14%.10 surveyed companies, however, fails to reveal extensive
disruptive innovation. Instead, for now, businesses are
much more likely to engage in modest experimentation.
10 MIT Technology Review Insights

03 Deployment in 2023:
Learning in pool’s
shallow end

E
nterprises have a nearly universal interest in For example, Laurence Liew, director of AI innovation
generative AI. A majority (76%) of at AI Singapore, says certain tasks such as document
respondents report that their companies have summarization will become “part and parcel of work
used the technology. Among the rest, most that people may occasionally do” because major
(18%) say they plan to do so soon. Moreover, office productivity suites now integrate generative AI-
about seven in 10 say “finding ways to derive business powered functions. On the other hand, although now
benefit from generative AI is the single leading priority of easier, such activity is similar to past practice.
at least one corporate function at my company.”
Survey data is consistent with these observations. In
Too narrow a focus on specific statistics, however, 2023, most businesses engaged in careful, limited
obscures the broader picture. The business use rather than wholesale transformation. One-third
implications of generative AI depend on the breadth of respondents conducted at least one generative
and impact of deployment. So far, businesses AI trial but did not permanently adopt any tools (see
are largely engaged in learning and modest Figure 2). A slightly higher percentage (37%) deployed
experimentation. Current efforts rarely seek out generative AI in specific areas but have yet to do so
disruptive possibilities as much as they could at this more generally. Only 9% of executives indicate their
stage. “While most businesses are exploring generative companies used generative AI widely. Even in sectors
AI capabilities, it is disappointing how tactical that that rely heavily on IT, adoption has only recently
experimentation has been,” says Schrage. “Too many started to spread. For example, just 20% of media
experts look at generative AI as a way of automating or and communications respondents and 18% of those
augmenting existing workflows and processes, rather in telecoms and technology report companywide
than rethinking use case fundamentals or the desired deployment.
outputs and outcomes they really want.”

“Too many experts look at generative AI as a way of

automating or augmenting existing workflows and
processes, rather than rethinking use case
fundamentals or the desired outputs and outcomes
they really want.”
Michael Schrage, Fellow, Center for Digital Business, MIT Sloan School
 MIT Technology Review Insights 11

Figure 2: The state of generative AI adoption

Which of the following best describes your company when it comes to adopting generative AI technology
across the business?
40%

35% 37%

30%
30%
25%

20%

15% 18%

10%

9%
5%
6%

0%
We have not adopted any We plan to do so, We have conducted We are using generative We are already
such generative but have not at least one trial. AI in certain, specific using generative AI
AI-powered applications or adopted any yet. areas and plan to deploy widely across the
platforms currently, and do the technology more organization.
not plan to do so. widely.

Source: MIT Technology Review Insights survey, 2024

“In Singapore, we are still in the very early stages,” innovation are media and communications (57%) and
says Liew. Even generative AI-enabled productivity technology and telecom (40%). Generative AI can
suites have only recently become available. Similarly, create text and video for media and PR. It can also
in Australia, Solar says “no holistic transformation or produce extensive code for specific purposes, boosting
approach that’s mature across companies” is yet visible. product development in these IT-centric fields.

The most frequent uses of generative AI focus on areas

where application is straightforward and unlikely to
disrupt industries. The most common, cited by 54%, is
automation of repetitive, low-level tasks. This appears
to be a gateway activity. Four out of 10 companies that
deployed generative AI for a single purpose in 2023 “In Singapore, we are still
used it in this way.
in the very early stages
The next tier of adoption consists of product innovation
(33%) and customer service (29%). These frequently [of using generative AI].”
reflect applications that draw on immediately
accessible strengths of generative AI. The sectors Laurence Liew, Director of AI Innovation,
that most frequently use the technology for product AI Singapore
12 MIT Technology Review Insights

0
Customer service is also looking to generative AI.
“Most organizations that deal with a large volume of “Across the market,
customers are really looking to figure out how to use
large language models to both power and become
generative AI is coming
platforms for next-generation chatbots,” says Schrage.
“Opportunities for personalization, segmentation, and
into organizations in
upselling here can only increase.” quite uncontrolled,
While inconsistent with the hype surrounding the non-centralized ways.”
transformational potential of generative AI, these
smaller efforts are important for the experience they Stela Solar, Inaugural Director, Australia’s
provide. Nuti says focusing on increasing productivity National Artificial Intelligence Centre,
and speed from available tools is a sensible “first
Commonwealth Scientific and Industrial
port of call” as a company grows to understand
Research
the potential benefits and risks of going further.
Moreover, there is substantial economic value in
these initial efforts. Pointing to the available tools that
UBS is adopting, Nuti says they “are going to make a
material difference to the productivity of UBS and our
employees.”
Most frequent users of generative AI for
The slow pace of generative AI adoption, however, product innovation
creates its own risks. Many employees want to work
with the same tools they have adopted personally.
“Across the market, generative AI is coming into
organizations in quite uncontrolled, non-centralized Media and
ways,” says Solar. “Employees are using it at work, and
most are not telling anyone.”
communications
A 2023 survey by Technology Decisions magazine in
Australia and New Zealand, for example, found that
63% are using AI in the workplace, but just 11% of their
organizations have a use policy. Technology
Given the risks, today’s levels of bring-your-own- and telecom
generative AI is dangerous. The survey, however,
shows this problem may be temporary. Adoption is
Source: Compiled by MIT Technology Review Insights survey, 2024
expected to accelerate markedly, reducing the need
for employees to use their own tools.
 MIT Technology Review Insights 13

04
I
2024: Year of
grand ambition

n contrast to 2023, plans for 2024 reflect intentions for a

full embrace of generative AI. Schrage says he sees the
first stirrings of the tech equivalent of an event known as
the “Cambrian explosion,” when a wide variety of animals
burst into the fossil record millions of years ago.
By the end of 2023, respondents used generative AI
for two purposes or fewer, on average. By the end of
2024, they expect that number to more than double
to exceed five. Just as notable, the technology will
proliferate in certain areas, particularly automation of
repetitive tasks, customer service, strategic analysis,
Innovation will occur around different kinds of products and product innovation (see Figure 3). More than 50%
and services, says Schrage. “Many of these are use of respondents in each industry say their companies will
case-driven,” he explains, while others are driven by what deploy generative AI in these fields. Most businesses
generative AI now allows companies to do. The survey (85%) say by the end of 2024 they will use generative
data indicates companies plan wide-ranging rollouts that AI to automate repetitive, low-value tasks that previously
could fuel an eruption of innovation. required considerable human effort.

Figure 3: How companies plan to use generative AI

In which areas are you currently using generative AI technology or will you begin in 2024?

Automating repetitive, low-value tasks 54% 31%

Customer service/experience 29% 48%

Strategic analysis 20% 54%

Product innovation 33% 41%

Supply chain logistics 13% 38%

Sales 12% 35%

Compliance 8% 35%

Coding/IT development 18% 24%

Risk/legal 7% 27%

None 22% 2%

Other 4% 3%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

Current Starting in 2024

Source: MIT Technology Review Insights survey, 2024
14 MIT Technology Review Insights

0
Generative AI-powered software will also become
the norm in customer service. Chatbots will become Generating product designs
increasingly ubiquitous as tools for agents, customers,
or both. Use of the technology, though, will go further. Calvin Wong, CEO, and center director of the
One example, says Schrage, is the development Laboratory for Artificial Intelligence in Design
of hybrid chatbots and wikis that provide “curated (AiDLab), which created the AI-based interactive
guidance” for customers who want detailed answers design assistant for fashion (AiDA), says the
and insights. Seller-buyer communication also looks traditional process of creating a new collection
set to deepen. Liew says a marketing company he for a fashion house typically involves designers
worked with has a new generative AI tool that allows creating a “mood board.” This consists of
customers to describe more precisely what they want diverse images including color tones, fabric print
the firm to produce, shortening the creative process. patterns, initial sketches, and possibly even the
upcoming season’s theme. In a time-consuming,
Meanwhile, only one in five companies used generative iterative process, designers use the board to
AI to support strategic analysis in 2023. But more than create design proposals, which are then refined
half (54%) plan to adopt it for this purpose in 2024, and finalized.
implying this area could see the largest spread of the
technology. Anecdotal evidence suggests that, once With AiDA, designers can upload their mood
properly trained and given a user-friendly interface, board to the platform. Within 10 seconds, the AI
generative AI can provide rapid insights on specific could generate thousands of suggestions. Based
questions. An INSEAD study combined the strategic on their judgement, designers can modify their
principles of the school’s marketing framework with proposals with AiDA’s suggestions, and feed
ChatGPT’s large language model (LLM) in this way. ideas back into AiDA to further explore them.
The resultant AI was directed to create a proposal for
a bagel bakery in Paris. It produced one in one hour: AiDA assists designers, but does not replace
the proposal was of similar quality to and, in some them. Creative control remains firmly in human
ways, more original than that made by MBA students, hands. The process, though, is dramatically faster
who took one week to complete the task.12 than the traditional method. It can cut the time
between initial proposal and completed collection
Product innovation is the last of the general areas by 70%, says Wong.
where respondents foresee widespread use of
generative AI by the end of 2024. AiDA, a generative
AI fashion design tool, shows what this might look like,
especially within creative industries (see sidebar). Just 8% of respondents from the highly regulated
financial services sector say they used generative AI
Beyond general use cases, specific industries plan to in compliance tasks in 2023. But 54% expect to begin
deploy generative AI in core operational areas in 2024. doing so in 2024. Nuti explains that while part of this
The survey shows nearly half (48%) of technology and involves accelerating existing tasks, the technology
telecom firms are already using it for coding and IT promises to achieve much more. For example, current
development. A further 32% expect to join them in 2024. regulations require banks to review a small sample of
written and telephone communication for conduct risk,
Meanwhile, most respondents (85%) from the logistics which is time consuming. Generative AI-enabled tools,
sector intend to use generative AI in supply chain however, can rapidly review every interaction.
management by the end of 2024. “A lot of supply-chain
leaders are looking for that great use case,” says an Schrage cites a financial services company that created
AI expert from a global IT consultancy. “It’s almost like a tool to answer compliance-related questions for
musical chairs. Everybody’s looking at what everybody developing new investment products. “The bet was that
else is doing.” Eventually, he expects managers to have it would lead to faster development and less painful
access to generative AI-enabled, complex digital twins, interactions with the compliance people when they went
giving them significant insights about their supply to market,” he says. “The organization in question felt
chains, suppliers, ecosystems, and customers. that they got real value for their time and effort.”
 MIT Technology Review Insights 15

05 Ambition and hubris

signal weak readiness
for rapid adoption

T
he majority of surveyed executives expect a The responses of those who rank the conduciveness
rapid rollout of generative AI in 2024 to of their companies’ IT capacities for rapid generative AI
support some key corporate activities. But adoption as between “not at all” and “modestly” reveal
organizations have a substantial agenda to weaknesses that could block a rapid and widespread
get through before they can fully embrace the rollout of generative AI (see Figure 5). More than half
technology. “There is a misconception about how easy it of surveyed executives rate their businesses poorly
is to run mature, enterprise-ready, generative AI,” says on hardware needed for generative AI adoption. More
Solar. “The large language model is almost the smallest than half also say the volume of data available for
part. There are surrounding elements like the app design, LLMs currently falls short.
connection to data and business processes, corporate
policies and more that are still needed.” These IT and Even these numbers are likely optimistic. Among
business challenges can stand in the way. “They often respondents, 9% say their companies have adopted
find a need to improve data quality and capability, privacy generative AI widely. The responses from these early
measures, AI skilling, and implement organization-wide adopters indicate technological weaknesses run deeper
safe and responsible AI governance,” she explains. than most executives fully realize (see Figure 6).

IT resources and capabilities often fall short

“Many organizations underestimate the requirements
for effective implementation of generative AI because
of the ease in using AI-enabled web pages or
consumer software,” says Solar. However helpful for
“There is a misconception
certain tasks, the impact of such tools will be limited.
Creating enterprise-level generative AI capabilities for
about how easy it is to run
competitive advantage is another matter. “You need
datasets, real datasets,” says Liew. “You need the AI
mature, enterprise-ready,
engineers who can actually build such applications and generative AI. The large
you need access to the computer infrastructure.”
language model is almost
the smallest part.”
Specialized IT assets that can effectively support
extensive, high-quality generative AI tools and
platforms are not yet widely deployed. Few
respondents, when asked to rank a range of relevant Stela Solar, Inaugural Director, Australia’s
IT-related capabilities and infrastructure elements, National Artificial Intelligence Centre,
indicate their companies’ assets are highly conducive Commonwealth Scientific and Industrial
for rapid generative AI adoption (see Figure 4). Research
16 MIT Technology Review Insights

Figure 4: Few perceive their IT assets for deploying generative AI as highly conducive
Percentage who consider the following highly conducive to the rapid adoption of generative AI at their
companies (on a scale of not at all conducive (1) to highly conducive (5)).

28% 27% 21% 20%

Quality/accuracy of available Ease of creating processes to Effectiveness of mitigating Ability to protect privacy of
data for large language models flag possible hallucinations cybersecurity risks customer/supplier data

19% 13% 7% 6%

Volume of data available for Data storage infrastructure Suitability of existing in-house Technical and business
large language models hardware and/or applicable considerations related to the
outsourced computing timely capture and analysis
platforms for generative AI of data

Source: MIT Technology Review Insights survey, 2024

“You need datasets, real datasets. You need the AI

engineers who can actually build such applications
and you need access to the computer infrastructure.”
Laurence Liew, Director of AI Innovation, AI Singapore
 MIT Technology Review Insights 17

Partner perspective

How Telstra is transforming itself into an AI-fueled organization

As generative AI has hit a critical mass, one of million scam calls and 11 million scam SMSs, and
the most profound technology revolutions of our almost 280 million incoming scam and potentially
time is set to transform the way we work and unwanted emails from reaching our customers
live. New opportunities are forged by disruption: each month.
from improving products through innovation to
exponentially enhancing the customer experience, Our work with a leading Australian bank on the
embracing AI is no longer optional. Scam Indicator is another prime example of
applying data and AI to detect high-risk scenarios
Telstra’s vision of becoming an AI-fueled organization and respond accordingly. Telstra created an API
is an ambition that doesn’t happen overnight. It’s a for the bank to integrate with its existing scam
multi-year journey that has required a fundamental detection processes, offering visibility on whether
shift: from considering and applying AI in isolated a customer is on a phone call—a key indicator
instances or on a use case basis, to embedding AI at they may be in the process of being scammed.
the heart of most decisions and processes.
The considerable benefits of AI are apparent but
Over the last few years, we have embarked on must be tempered with recognition of the risks.
a radical simplification and modernization of our And fervor can’t lead to taking shortcuts with
underlying technology architecture and data good governance.
ecosystem. We’ve shifted to a 100% application
programming interface (API)-first approach to From the outset, the responsible and ethical
product development. We’re migrating workloads to development and deployment of AI has anchored
the cloud and building reusable AI capabilities. our approach. It’s why we’ve consulted and
co-developed with the Australian government
We are now using AI to improve half of our key and tech leaders across the globe to develop
processes, including to automatically detect and strict guardrails based on human-based values.
resolve fixed services faults, and to solve customer Principles such as privacy, security, contestability,
issues faster. For example, we have piloted new and accountability are absolutely critical to
AI applications including Ask Telstra—an OpenAI- leveraging AI in a responsible manner and are
based solution to help our frontline teams find the supported by other robust frameworks and
information they need to serve our customers better controls.
and more quickly.
The path from early adoption to becoming truly
We’re also investing in our people, including through AI-fueled will be one that requires an unwavering
our Data & AI Academy, to upskill them in AI and help conviction, but an agility allowing us to flex to
them understand how they can use it in their roles. seize opportunities as they arise.

Cybersecurity, identity, and scam protections remain What lies ahead is unknown but as we accelerate
extremely important to us and our customers. our adoption of AI, what is clear is that it will
Through our Cleaner Pipes initiative1, as of February require a paradigm shift and an infusion of data
2024 we are blocking on average more than 10 and AI into the very fabric of our company.

1. Source: https://www.telstra.com.au/content/dam/tcom/about-us/investors/pdf-i/ceo-cfo-analyst-briefing-presentaion-and-materials.pdf
18 MIT Technology Review Insights

Figure 5: Overview of IT assets that fall short for rapid generative AI adoption
How conducive are the following to the rapid adoption of generative AI tools at your company?
(Percentage answering between “not at all” and “modestly”)

Volume of data available for large language models 56%

Suitability of existing in-house hardware and/or applicable

51%
outsourced computing platforms for generative AI
Technical and business considerations related to the timely 41%
capture and analysis of data

Data storage infrastructure 41%

Effectiveness of mitigating cybersecurity risks 36%

Ability to protect privacy of customer/supplier data 34%

Quality/accuracy of available data for large language models 29%

Ease of creating processes to flag possible hallucinations 27%

(e.g. human or technological review of output before usage)
0% 10% 20% 30% 40% 50% 60%
Source: MIT Technology Review Insights survey, 2024

Figure 6: Early adopters rate the conduciveness of their IT assets and capabilities harshly
How conducive are the following to the rapid adoption of generative AI tools at your company?
(Percentage answering between “not at all” and “modestly”)

Suitability of existing in-house hardware and/or applicable 65%

outsourced computing platforms for generative AI 50%

58%
Volume of data available for large language models 56%

Effectiveness of mitigating cybersecurity risks 54%

34%

Quality/accuracy of available data for large language 50%

models 27%

50%
Data storage infrastructure 41%

Ease of creating processes to flag possible hallucinations 38%

26%

Ability to protect privacy of customer/supplier data 31%

35%

Technical and business considerations related to the 27%

timely capture and analysis of data 43%

0% 10% 20% 30% 40% 50% 60% 70%

Generative AI widely deployed Other respondents

Source: MIT Technology Review Insights survey, 2024
 MIT Technology Review Insights 19

“Building end-to-end Calvin Wong, CEO and center director of Hong Kong-
based Laboratory for Artificial Intelligence in Design

capabilities to handle (AiDLab), says clients’ inability to provide a large

dataset is the biggest barrier for his organization’s
large datasets, accurately generative-AI fashion design tool. “Data is the key
component if proposals generated by generative AI
contextualize the data for tools are to meet the actual requirements of a specific
industry,” he says. “The dataset is a common problem.”
business value and ensure
Perhaps, as a result, companies are beginning to
minimal AI hallucinations rethink their approach to data. “Generative AI has

is extremely challenging.” turned on its head the value of data in an organization,”

says Solar. Nevertheless, firms need to scrutinize
their data practices. “They have to consider what
Geraldine Kor, Managing Director of South data they actually need to add value on top of existing
Asia and Head of Global Enterprise, Telstra foundation models,” she adds. “Are they collecting
International the right data? Do they need to keep all of that data
because data, as well as being valuable, also adds
exposure to cybersecurity risks.”
Deploying generative AI across the company seems to
douse tech optimism. Nearly two-thirds of early adopters
Geraldine Kor, managing director of south Asia and
acknowledge their available hardware is at best modestly
head of global enterprise at Telstra International, says
conducive to rapid adoption. Half or more say the same
“in today’s hyperconnected business and AI landscape,
about the volume (54%), quality (50%), and storage
it is essential that IT leaders embark upon a technology
infrastructure of their data (50%). In almost every case,
adoption strategy that right-fits, right-sizes and right-
those who have implemented generative AI widely are
locates their IT investments.” Determining the optimal
more likely to rate their IT more harshly. Problems with
deployment scenarios for storing, analyzing, and
hardware and data are particularly notable.
interconnecting large data volume sets and business
logic that can meet the latency-sensitivity of modern
Hardware: Appropriate hardware, in-house or
applications can be extremely complex, she adds.
outsourced, is a prerequisite of extensive generative AI
adoption, and executives often fail to grasp the degree
of the requirement.

Accessing these assets poses a dilemma. Outright

purchase carries risk and in a fast-moving landscape,
over-committing is risky. “Rather than building a
“Data is the key
massive data center, it might be more appropriate to
remain nimble, keep an eye on developments, and try to
component if proposals
do as much pay-as-you-go as you can,” says Nuti. Many generated by generative
large suppliers can be found for such outsourcing, adds
Liew, but this solution brings its own risks. “You may not AI tools are to meet the
necessarily have access to this infrastructure when you
want it,” he says. actual requirements of a
Datasets and volume: Also in short supply—and specific industry.”
fundamental for generative AI—is enough good data.
Among early adopters, 54% say available volume of Calvin Wong, CEO and Center Director,
data at their companies is at best modest, and half say Laboratory for Artificial Intelligence in
the same of data accuracy and storage infrastructure. Design
20 MIT Technology Review Insights

Data quality and storage: Even though generative

AI is better at using unstructured data than other Need for a broader data
technologies, data quality and storage remain basic
requirements for effective deployment. “When you
strategy
are doing the fine-tuning, you actually need cleaner
Amid the specific data-related requirements
datasets,” says Liew, referring to LLMs and training
made more pressing by the advent of generative
them to respond to prompts. More generally, he adds,
AI, companies need to embed their improvements
appropriately “structured datasets still need to be
in a broader data strategy.
curated and collected.” Any deficiency in this area
will slow adoption. For supply chain companies, data
“As the world becomes increasing digitized
is the biggest barrier because so much information
and human-to-machine interactions flourish,
in the industry has traditionally lived on spreadsheets
being able to process data to drive informed
and siloed databases differentiated by application or
real-time or near real-time business decisions
location, says an AI expert at a global IT consultancy.
is paramount”, says Geraldine Kor, managing
director of south Asia and head of global
Non-technological barriers pose daunting,
enterprise at Telstra International.
long-term challenges
Looking beyond IT, respondents identified other
“When implemented successfully, this proficiency
barriers or enablers for rapid adoption of generative
will be a game-changer for most organizations,
AI. The regulatory, compliance, and data privacy
and will distinguish leaders from followers,” adds
environment was the most commonly mentioned
Kor. “However, building end-to-end capabilities to
impediment (see Figure 7). More than half of early
handle large datasets, accurately contextualize
adopters say they struggle to address cybersecurity
the data for business value and ensure minimal AI
(see Figure 6).
hallucinations is extremely challenging.”

“Legal regimes, not technological capabilities, pose the

Dayle Stevens, Telstra’s data and AI executive
biggest threat to the development and deployment of
agrees. “A broad, well-thought-out data strategy
generative AI,” says Schrage. Regulators and litigators
is pivotal for leveraging generative AI effectively,”
could rein in businesses that overstep existing rules
she says. “It underpins ethical AI use, ensures
with the new technology.
regulatory compliance, and prepares the
organization for future technological shifts. This,
in turn, positions companies to harness AI’s full
potential, driving innovation and competitive
advantage.”

“It’s still unclear whom to

hold accountable for There are myriad problems for companies in staying
on the right side of this issue. For example, Solar
what, in order to create argues, the challenge that generative AI creates
in differentiating truth and falsehood is already
safe and responsible AI disrupting the cybersecurity threat landscape.

outcomes.”
Similarly, unsanctioned employee use of generative AI
undermines essential risk assessment processes and
corporate policy implementation.
Stela Solar, Inaugural Director, Australia’s
National Artificial Intelligence Centre, Just as problematic, Solar adds, is lack of clarity on
Commonwealth Scientific and Industrial accountability. “An AI solution has many vendors,
Research providers, services, and stakeholders that contributed
 MIT Technology Review Insights 21

“A broad, well-thought-out data strategy is pivotal for

leveraging generative AI effectively.”
Dayle Stevens, Executive – Data & AI, Telstra

towards it: data that’s bought from somewhere; a Rather than adopt and hope, companies must place
cloud platform that’s purchased; machine learning risk governance at the center of the technology’s
toolkits; advisory that’s provided,” she says. “It’s still adoption, says Nuti. UBS’s generative AI governance
unclear whom to hold accountable for what, in order process gives its risk department “full visibility,
to create safe and responsible AI outcomes.” This understanding, and the ability to control the risks of
is particularly worrying as, in the race to create AI these new applications,” he adds.
technologies, some development teams at startups
and even mature companies have completely Others use different but comprehensive approaches.
bypassed internal regulatory and compliance Solar says many organizations are pursuing a cohesive
checkpoints, she says. combination of “implementing or endorsing a set of
approved generative AI tools, establishing generative
For UBS, risk and compliance are the biggest AI policies, and proactive employee upskilling.” She
non-technological challenges for generative AI has also seen a new role—the “trust architect”—begin
adoption. “Potentially, many risks come with this new to appear. These professionals “stitch technology
technology that we need to ponder carefully,” says processes, business processes and siloed company
Nuti. These may be heightened versions of existing systems together, to ensure that safe and responsible
ones, such as data privacy, or entirely new dangers. outcomes are supported throughout the system.”

Figure 7: Non-IT barriers to rapid generative AI adoption

Which of the following are the three leading enablers/drivers of rapid adoption of generative AI at your
company and which are the top three barriers?

77%
Regulatory/compliance/data privacy environment 10%

Budget for technology investment 56%

30%

Competitive environment 51%

25%

Cultural attitudes to technological innovation within the 41%

company 24%

Pool of relevant technical talent to hire within the 25%

domestic economy 47%

Level of relevant technical skills within the company 22%

54%

Level of business skills to apply insights from new 14%

generative AI-powered applications within the company 54%

Leadership attitudes/understanding of possible benefits 10%

54%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90%

Barriers Enablers
Source: MIT Technology Review Insights survey, 2024
22 MIT Technology Review Insights

Dayle Stevens, data and AI executive at Telstra, equivalent figures are 23% and 54%. Culture is another
argues these policies provide guardrails to develop differentiator: 50% of the early adopters call culture a
and deploy AI responsibly and are non-negotiable. “AI driver of rapid AI adoption and 31% say it is a barrier.
possesses immense potential to deliver benefits but Within the rest of the respondents, 22% say culture is
create a minefield of risks,” says Stevens. “Robust a driver, and 42% call it a barrier.
checks and balances have to be included by design
from the outset to ensure principles, based on human- Cultural reticence can undermine the benefits of
centered values, are observed. As AI rapidly scales, generative AI adoption. Wong says some fashion
we all must have a deep commitment to its ethical designers trained in conventional design processes
application, which should be anchored to goals such for years are reluctant to adopt the AiDA tool. Even
as privacy, security, contestability, transparency, and though many appear excited about generative AI
explainability.” before it is deployed, they show concern when it
becomes real, he says. They seem worried about
The second most common barrier reported by losing their jobs.
respondents are budgetary constraints. The numerous
technological challenges for generative AI adoption Company culture is not immutable. Wong believes it
indicate that success may require substantial is essential to educate users and executives in a way
investment. This problem is compounded at firms that “puts the focus on the interaction between human
that focus too much on cost when mulling spending creativity and generative AI.” They can then see that
decisions. “Look for business cases around growth, the technology assists rather than replaces them.
margin, cost reduction, and process efficiency, and Openness to building trust is also essential. “Many
then decide what you are prepared to spend to business leaders will start by making decisions that
leverage generative AI,” Schrage says. Unfortunately, play safe in a window of normalcy,” says an AI expert
he adds, this appears still to be a minority view. at a global IT consultancy. “When these initial pilot
projects come up with good results, then trust will
Early adopters see things differently grow, and leaders will be more comfortable expanding
The third and fourth most frequently mentioned their use cases.”
barriers to generative AI adoption (see Figure 7)
are the competitive environment (cited by 51% Early adopters and other surveyed executives also
of respondents) and cultural attitudes toward differ around talent. Non-early adopters are more
technological innovation (41%). Early adopters, optimistic: a minority (21%) of them say the skill levels
however, see both factors as drivers of rapid adoption within their workforce are a barrier to rapid adoption.
rather than impediments. A similar number (23%) say the same of skills within
their domestic economies. Meanwhile, 31% of early
Almost half of early adopters (46%) call the adopters say skills within their own companies are a
competitive environment an enabler, while two in barrier, while 46% say the same of available external
10 call it a barrier (19%). For other respondents, the talent.

“As AI rapidly scales, we all must have a deep

commitment to its ethical application, which should be
anchored to goals such as privacy, security,
contestability, transparency, and explainability.”
Dayle Stevens, Executive – Data & AI, Telstra
 MIT Technology Review Insights 23

Figure 8: Shortage of AI talent

Share of respondents reporting difficulty in organizations’ hiring of AI-related roles (%)

Less difficult More difficult

Source: Compiled by MIT Technology Review Insights from McKinsey & Company, 2024

Talent is a bottleneck for many firms wishing

to build their own LLMs, says Liew. Nuti says,
beyond IT skills, even those who use generative
AI tools need the right knowledge and skillsets to
understand its risks and potential pitfalls.

Others say it is hard to find the talent to build

generative AI platforms for business uses, due
to the state of the market. Accordingly, Solar
notes, many organizations are engaged in skilling
efforts. Eventually, supply and demand may solve
the problem. A McKinsey 2023 survey of more
than 1,600 participants found that, between 2022
and 2023, the number of companies reporting
difficulties in hiring relevant technical skills
declined, albeit to still high levels (see Figure 8).
24 MIT Technology Review Insights

06 Conclusion: Grand plans

or unrealistic dreams

G
enerative AI is a technological advance Given the potential competitive value of generative AI,
of huge importance. It could irrevocably many companies are likely to pursue an aggressive
reshape key business operations and adoption strategy. The question in 2024 is how
enable extensive innovation. Senior many businesses are ready to deploy generative AI
executives from various industries in this effectively. Survey responses indicate many are in
survey expect adoption of generative AI to cause danger of leaving themselves open to being disrupted
substantial disruption. They primarily see the rather than becoming disruptors.
opportunity to secure competitive advantage as the
technology becomes more pervasive and continues to
mature.

Their actions, however, paint a less ambitious picture.

Few companies rolled out generative AI widely in
2023. It is used most commonly for automation of
low-value tasks, rather than for innovative use cases or
reshaping business models. This only begins to tap its
potential.

Polled executives appear to agree broadly that a

wider rollout is coming. In every surveyed industry, the
majority of respondents hope to explore or expand
use of generative AI in customer relations, strategic
analysis, and product innovation by the end of 2024.
Specific sectors will likely embark upon wider adoption
of generative AI in core areas, such as supply chain
management for logistics and compliance at financial
services firms.
 MIT Technology Review Insights 25

The question in 2024 is how many businesses are ready

to deploy generative AI effectively. Survey responses
indicate many are in danger of leaving themselves open
to being disrupted rather than becoming disruptors.
The survey uncovered specific challenges companies Competitive environment: Generative AI for its own
should be mindful of as they seek the opportunities sake makes limited business sense. Early adopters
presented by generative AI. have learned that companies need to deploy it
specifically to improve competitive position.
IT infrastructure: Respondents, especially early
adopters, have reservations about whether their Culture: Effective change management requires
companies’ hardware can provide the backbone for openness to innovation. Early adopters say it is also a
generative AI platforms and tools. The choices are key factor for rapid generative AI adoption.
complex and require planning.
Skills: The IT skills to run generative AI projects are in
Data: Generative AI requires large volumes of high- short supply. Those using generative AI-powered tools
quality, accessible data. A coherent strategy is need to understand the risks, limitations, and potential
essential. Firms that have not thoroughly digitalized benefits. Talent acquisitions and development will
both the front- and back-end of their businesses have require investment.
another pressing reason to do so.
Disruption can happen to any company. But bringing
Risk management: Regulatory, compliance, data, and about that disruption requires the adoption of tools and
cybersecurity risks have multifaceted implications attitudes in the new competitive environment that they
that create the biggest non-technological barrier to are seeking to create.
generative AI deployment. Any strategy to use the
technology should be rooted in holistic risk governance
structures.

Budgets: Generative AI can be expensive. In a

marketplace poised for substantial disruption,
executives should examine use cases in terms of ROI,
not just implementation costs.
26 MIT Technology Review Insights

Endnotes

1 This report focuses specifically on generative AI, one of the critical technologies that fall under the broader category of emergent AI. Gartner has highlighted emergent AI
as one of four hype cycle themes to think about in 2023 and beyond (see “What’s New in the 2023 Gartner Hype Cycle for Emerging Technologies,” August 2023, https://
www.gartner.com/en/articles/what-s-new-in-the-2023-gartner-hype-cycle-for-emerging-technologies).

2 Michael Chui, Eric Hazan, Roger Roberts, Alex Singla, Kate Smaje, Alex Sukharevsky, Lareina Yee, and Rodney Zemmei, “The economic potential of generative AI: The next
productivity frontier,” McKinsey, June 14, 2023, https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-
productivity-frontier.

3 “Generative AI to Become a $1.3 Trillion Market by 2032, Research Finds” (press release), Bloomberg, June 1, 2023, https://www.bloomberg.com/company/press/
generative-ai-to-become-a-1-3-trillion-market-by-2032-research-finds/.

4 Ingrid Lexova, “Risks, regulation in focus as AI boom accelerates,” S&P Global Market Intelligence, December 4, 2023, https://www.spglobal.com/marketintelligence/en/
news-insights/latest-news-headlines/risks-regulation-in-focus-as-ai-boom-accelerates-79585293.

5 Jasmine Cui, “There’s a gap between AI talk and companies using it,” December 4, 2023, https://www.nbcnews.com/data-graphics/wide-gap-ais-hype-use-business-
rcna127210.

6 Lori Perri, “What’s New in the 2023 Gartner Hype Cycle for Emerging Technologies,” August 23, 2023, https://www.gartner.com/en/articles/what-s-new-in-the-2023-
gartner-hype-cycle-for-emerging-technologies.

7 “Gartner’s 2015 Hype Cycle for Emerging Technologies Identifies the Computing Innovations That Organizations Should Monitor” (press release), Gartner, August 18, 2015,
https://www.gartner.com/en/newsroom/press-releases/2015-08-18-gartners-2015-hype-cycle-for-emerging-technologies-identifies-the-computing-innovations-that-
organizations-should-monitor.

8 Michael Chui, Eric Hazan, Roger Roberts, Alex Singla, Kate Smaje, Alex Sukharevsky, Lareina Yee, and Rodney Zemmei, “The economic potential of generative AI: The next
productivity frontier,” McKinsey, June 14, 2023, https://www.mckinsey.com/capabilities/mckinsey-digital/our-insights/the-economic-potential-of-generative-ai-the-next-
productivity-frontier.

9 Sida Peng et al., “The Impact of AI on Developer Productivity: Evidence from GitHub Copilot,” arXiv.org, February 13, 2023, https://arxiv.org/pdf/2302.06590.pdf.

10 Erik Brynjolfsson et al., “Generative AI at Work,” NBER Working Paper No. 31161, 2023, April 2023, https://www.nber.org/system/files/working_papers/w31161/w31161.pdf.

11 Dylan Bushell-Embling, “Shadow AI use in the workplace widespread: survey,” Technology Decisions, November 16, 2023, https://www.technologydecisions.com.au/content/
it-management/news/shadow-ai-use-in-the-workplace-widespread-survey-1152526171.

12 Michael Olenick and Peter Zemsky, “Can Gen AI Do Strategy?” Harvard Business Review, November 24, 2023, https://hbr.org/2023/11/can-genai-do-strategy.
 MIT Technology Review Insights 27

About MIT Technology Review Insights

MIT Technology Review Insights is the custom publishing division of MIT Technology Review, the world’s
longest-running technology magazine, backed by the world’s foremost technology institution—producing
live events and research on the leading technology and business challenges of the day. Insights conducts
qualitative and quantitative research and analysis in the US and abroad and publishes a wide variety of content,
including articles, reports, infographics, videos, and podcasts. And through its growing MIT Technology Review
Global Insights Panel, Insights has unparalleled access to senior-level executives, innovators, and
entrepreneurs worldwide for surveys and in-depth interviews.

From the sponsor

Telstra is a leading telecommunications and technology company with a proudly Australian heritage and a
longstanding, growing international business. Today, we operate in over 30 countries outside of Australia,
providing services to thousands of business, government, carrier, and OTT customers.

Telstra Enterprise is a division of Telstra that provides data and IP networks and network application services,
such as managed networks, unified communications, cloud, industry solutions and integrated services.

Telstra Purple, our professional and managed services business in Australia, Asia, and the UK, brings together
people and innovative solutions to define and deliver a clear vision of our customers’ transformation journey,
network foundation, and the protection they need to thrive.

From managed adaptive networks to securing data across international supply chains or empowering people to
work from anywhere, our over 1,500 certified experts offer innovative solutions to help open and develop new
markets for businesses.

For more information, visit www.telstra.com.sg.

Illustrations
Illustrations assembled by Tim Huxford with elements from Shutterstock and Adobe Stock.

While every effort has been taken to verify the accuracy of this information, MIT Technology Review Insights cannot accept any responsibility or liability for reliance by any person
on this report or any of the information, opinions, or conclusions set out in this report.

© Copyright MIT Technology Review Insights, 2023. All rights reserved.

MIT Technology Review Insights
www.technologyreview.com
[email protected]