text
stringlengths 0
182
|
|---|
years to become integrated into business productivity
|
tools.5 To an extent, the febrile current discussion
|
around generative AI, however, obscures some
|
attributes that will shape its impact.
|
First, generative AI is a rapidly evolving field: it did not
|
spring into existence with the release of ChatGPT
|
and its development is far from complete. McKinsey
|
describes a rush to throw money at all things generative
|
AI between 2017 and 2022. During those years, private
|
investments in the technology rose at an average annual
|
compound rate of 74%, far outstripping the equivalent
|
growth rate for AI as a whole of 29%.6 Similarly, tools and
|
products incorporating generative AI have continued to
|
develop rapidly in recent months (see Figure 3).
|
Second, generative AI looks set to build on and
|
supplement existing technologies, rather than
|
necessarily replace them on a grand scale. As Chia
|
notes, you definitely need a clear use case to start
|
trying out any new technology. Any such cases will
|
arise from what generative AI can do compared to
|
existing tools.
|
These may be more limited than popular imagination
|
appreciates. Tools enabled by older versions of AI,
|
for example are now called just software, says John
|
Mileham, chief technology officer at Betterment, a
|
robo-advisor that assists users in automated investing.
|
And such tools have already played a crucial role in
|
the ongoing digitalization of financial services firms
|
and other companies. Theres little point in deploying
|
generative AI where less advanced technologies are
|
doing the job as effectively and at low cost. Indeed,
|
some analysts project that, while generative AI will have
|
a very large economic impact, it will be markedly less
|
than that of earlier iterations of AI.7
|
Its not magic
|
The use cases will arise from what generative AI
|
can deliver that other tools cannot. The striking new
|
capacity of generative AI is best illustrated by a brief
|
Figure 2: The breakout technology of 2023
|
ChatGPT gained more than 100 million active users across the globe within a span of two months, a much
|
faster rate than that of other platforms.
|
Source: Compiled by MIT Technology Review Insights, based on data from Generative Artificial Intelligence in Finance: Risk Considerations, International
|
Monetary Fund, 2023.
|
70
|
60
|
50
|
40
|
30
|
20
|
10
|
0
|
ChatGPT
|
TikTok
|
Uber
|
Instagram
|
Telegram
|
Pinterest
|
Facebook
|
Spotify
|
Twitter
|
Months to gain 100 million users
|
8
|
MIT Technology Review Insights
|
comparison with older forms of AI. Put simply, older
|
AI tools can train themselves on huge amounts of
|
structured data and can answer specific questions
|
asked in programming-like language.
|
Generative AI can learn from an even larger body of
|
informationincluding unstructured dataand it can
|
create apparently novel content in response to natural
|
language questions. Older AI tools, for example, can
|
tell users if something is a cat. Generative AI can
|
generate a new image of a cat.
|
In short, generative AIs strength is that it allows users
|
to ask questions in natural language and receive
|
output that provides readily comprehensive answers
|
in different formats based on a huge corpus of
|
information. This can involve, among other things, the
|
generation of new text, pictures, computer code, or
|
even large data sets.
|
The potential value of this creativity is substantial but
|
should not be overestimated. For example, generative
|
AI can generate a data set based on existing ones, says
|
Rama Cont, chair of mathematical finance and head of
|
the Oxford mathematical and computational finance
|
group at Oxford University. However, while it may be
|
even 100 times larger, it wont have more information,
|
he explains. It can extrapolate to certain situations,
|
provided you have similar data sets on which to train it.
|
In the cat example, the picture will rely on existing
|
knowledge of cats but will be unable to create something
|
new. Such extrapolation of existing data is a nice
|
feature of generative AI, says Cont, but not magic.
|
Figure 3: Six months in the life of generative AI
|
Starting in late 2022, new iterations of generative AI technology have been released several times a month.
|
Timeline of major large language model (LLM) developments following Chat GPTs launch
|
Nov. 2022
|
Dec.
|
Jan. 2023 Feb. March April
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.