人工智能技术的发展和应用前景

版权申诉

191 浏览量更新于2024-07-03 收藏 3.27MB PDF 举报

人工智能概述人工智能（Artificial Intelligence）是信息时代的顶级技术。随着机器学习和深度学习的进步，结合更加强大的计算能力和不断扩展的数据池，人工智能正在逐步走近公司跨行业的应用范围内。人工智能作为服务的发展拥有打开新市场和打破云计算领域格局的潜力。我们认为，能够充分利用人工智能的能力将成为公司未来的竞争优势关键属性，并将带来生产率的复苏。以下是人工智能概述的相关知识点：一、人工智能定义人工智能是指通过机器和计算机系统来模拟人类智能，实现自动化、智能化和自适应的技术。人工智能包括机器学习、深度学习、自然语言处理、计算机视觉等领域。二、人工智能的应用场景人工智能的应用场景非常广泛，包括但不限于： * 客户服务：人工智能可以应用于客户服务领域，例如智能客服、自动化客服等。 * 数据分析：人工智能可以应用于数据分析领域，例如数据挖掘、预测分析等。 * 图像识别：人工智能可以应用于图像识别领域，例如人脸识别、物体识别等。 * 自动化生产：人工智能可以应用于自动化生产领域，例如智能制造、自动化装配等。三、人工智能的技术架构人工智能的技术架构主要包括： * 数据层：数据层是人工智能的基础，负责数据的采集、存储和管理。 * 算法层：算法层是人工智能的核心，负责算法的设计和实现。 * 应用层：应用层是人工智能的最终应用，负责将算法应用于实际场景中。四、人工智能的挑战和机遇人工智能面临着许多挑战，例如数据隐私、算法公平性、工作岗位替代等。但是，同时人工智能也提供了许多机遇，例如提高生产率、改善生活质量等。五、人工智能的发展趋势人工智能的发展趋势包括： * 云计算人工智能：云计算人工智能是指将人工智能技术应用于云计算领域，提供更加智能和自动化的云计算服务。 * 边缘人工智能：边缘人工智能是指将人工智能技术应用于边缘计算领域，提供更加快速和智能的边缘计算服务。 * Explainable AI：Explainable AI是指能够解释人工智能决策过程的技术，提高人工智能的透明度和可靠性。人工智能是信息时代的顶级技术，具有广泛的应用场景和技术架构。人工智能的发展趋势包括云计算人工智能、边缘人工智能和Explainable AI等。

November 14, 2016 Profiles in Innovation

Goldman Sachs Global Investment Research 16

operations processes contributed to creating a threefold spike in growth (output per labor

hour) during the 1990s, with IT-producing sectors responsible for at most 49% of the

average annual increase in annual productivity growth from the pre-boom period to the

period between 1995 and 2004 (Exhibit 10).

Exhibit 10: Late 90s: IT-producing sectors contribute nearly half of productivity growth

But lose value and share in growth post-tech boom

Source: Federal Reserve Board, Goldman Sachs Global Investment Research

Post-millennium stagnation. During the past decade, capital deepening growth related to

IT applications (computer hardware, software, and telecom) has stagnated. IT capital,

relative to broader market capital, has contributed less to overall growth in this component

than average contributions during and even before the tech boom. Aggregate labor hours

have been increasing, but the contribution of capital intensity to productivity has drastically

underperformed versus the 1990s. The introduction of increasingly sophisticated,

consumable machine learning and AI may be a catalyst in bringing capital intensity back to

the forefront, in our view, significantly increasing the productivity of labor similar to the

cycle we saw in the 1990’s.

We’re more optimistic on the MFP side of the equation. GS economists have highlighted

(Productivity Paradox v2.0 Revisited, 9/2/2016) that upward biases on ICT prices and a

growth in inputs towards unmonetized outputs (free online content, back-end processes,

etc.) add to the understatement of real GDP and productivity growth. Evolution of internet

giants like Facebook and Google highlight the idea that complex input labor and capital

aren’t necessarily converted into traditional consumer product monetization captured in

standard productivity metrics.

AI/ML induced productivity could impact investment

We believe that one of the potential impacts of increasing productivity from AI/ML could be

a shift in the way companies allocate capital. Since mid-2011, the growth in dividends and

share repurchases has significantly exceeded capex growth, as reluctance among

management teams to investment in capital projects remains post-recession.

0.77

1.5

0.64

0.79

1.56

0.92

1974-1995 1995-2004 2004-2012

IT Contribution Other nonfarm business

% growth

1.56% total average

growth virtually equal

to 1995-2004 average

IT contribution

via：资料来源网络，起点学院学员收集

起点学院www.qidianla.com，互联网黄埔军校，打造最专业最系统的产品、运营、交互课程。

November 14, 2016 Profiles in Innovation

Goldman Sachs Global Investment Research 18

GS Research Internet analyst Heath Terry sat

down with Chief Economist Jan Hatzius to

discuss the role AI & machine learning could

play in boosting lagging labor productivity.

Heath Terry: W

hat has led to the lack of

measurable productivity growth over the

last decade?

Jan Hatzius:

A good starting point is the

1990s, where we did have a sizeable

measured productivity acceleration which

was mainly technology driven. The

technology sector had gotten bigger and

measured output growth in the technology

sector was very rapid, and that was enough to produce an overall

acceleration in the economy. Recently, over the last 10 years or so,

we’ve seen a renewed deceleration to productivity growth rates

that are as low as the 1970s and 1980s, and potentially even lower.

I think there is more than one driver, but I would say there are three

things that have lowered productivity growth in my view. One is a

bit of a cyclical effect. We still have some hangover from the great

recession with a relatively slow pace of capital accumulation,

relatively low levels of investment, and relatively rapid employment

growth. Since productivity is measured as output per hour worked,

that somewhat perversely means that you can have lower

productivity numbers when the labor market is improving rapidly.

Another factor may be some slowdown in the overall pace of

technological progress. It’s reasonable to believe that perhaps the

1990s, with the introduction of the internet, was a relatively rapid

period for technological progress, and I would say there is some

support for the idea that it is a little slower now.

The third point is that technological progress that has occurred

over the last decade, like mobile and consumer focused

technology, is something that statisticians are not very well set up

to capture in the numbers. The reason is that quality improvement

in a lot of the new technologies that have dominated in the last

decade or so is very difficult to capture in a quantitative sense.

Statisticians are not building in significant quality improvement

into a lot of the areas that have been at the cutting edge.

So I would point to three things. There are cyclical effects,

probably some slowdown in technological progress, and very

likely an increase in the measurement error in the productivity

statistics.

Terry: Back to the productivity boom in the 1990s, what role did

technology play?

Hatzius:

What drove it was mainly general purpose technologies

like semiconductors and computers, which had become much

larger as a share of the economy than they were in the 1970s or

1980s and where technological progress was very rapid, in ways

that statisticians were well set up to measure. The statisticians in

the 1990s had made a lot of effort to capture quality improvement

faster; processer speeds, more memory, better attributes in

computer hardware, which led to large increases in the measured

contribution of the technology sector. The technology sector was

very central to pick up in the productivity numbers from the 1990s

lasting to the early and mid-2000s.

Terry:

We’ve seen a lot of technology development over the last

10-15 years. Why hasn’t there been a similar impact to

productivity from technologies such as the iPhone, Facebook, and

the development of cloud computing?

Hatzius:

We don’t have a full answer to it, but I do think an

important part of the answer is the statistical ability to measure

improvement in quality, and the impact of new products in the

economic statistics is limited. It’s relatively easy to measure

nominal GDP, that’s basically a matter of adding up receipts. There

is room for measurement error as there is in almost everything,

but I don’t have real first order concern that measurement is

getting worse in terms of measuring nominal GDP. Converting

nominal GDP numbers into real GDP numbers by deflating it with a

quality adjusted overall price index is where I think things get very

difficult. If you look, for example, at the way software sectors enter

the official numbers, if you believe the official numbers, $1000 of

outlay on software now buys you just as much real software as

$1000 of outlay bought you in the 1990s. There has been no

improvement in what you get for your money in the software

sector. That’s just very difficult to believe. That just does not pass

the smell test. Because of the difficulty of capturing these quality

improvements, the fact that the technology sector has increasingly

moved from general purpose hardware to specialized hardware,

software and digital products has led to increased understatement

and mismeasurement.

Terry:

What kind of impact could the development of technologies

like artificial intelligence and machine learning have on

productivity?

Hatzius: In principle, it does seem like something that could be

potentially captured better in the statistics than the last wave of

innovation to the extent that artificial intelligence reduces costs

and reduces the need for labor input into high value added types of

production. Those cost saving innovations in the business sector

are things statisticians are probably better set up to capture than

increases in variety and availability of apps for the iPhone, for

example. To the extent Artificial Intelligence has a broad based

impact on cost structures in the business sector, I’m reasonably

confident that it would be picked up by statisticians and would

show up in the overall productivity numbers.

I would just add one general caveat, which is that the U.S.

economy is very large. Even things that are important in one sector

of the economy and seem like an overwhelming force often look

less important when you divide by $18tn, which is the level of U.S.

nominal GDP, so the contribution in percentage terms may not

look as large as one might think from a bottom up perspective. But

in principle, this is something that could have a measurable

impact.

Terry: You touched on the impact to cost, how do you see

something like that impacting pricing? Does that become a

contributor to the broader deflationary force that we’ve seen in

certain parts of the economy?

Hatzius: I certainly think that in terms of productivity gains, the first

order impact is often to lower costs and lower prices. Keeping

everything else constant would mean lower overall inflation in the

economy. It’s often not the right assumption that you want to keep

AI & The Productivity Paradox: An interview with Jan Hatzius

via：资料来源网络，起点学院学员收集

起点学院www.qidianla.com，互联网黄埔军校，打造最专业最系统的产品、运营、交互课程。

November 14, 2016 Profiles in Innovation

Goldman Sachs Global Investment Research 19

everything else constant; there are economic policy makers and

there is a Federal Reserve, and if the impact is large then the

Federal Reserve is going to run an easier monetary policy and

allow people who may no longer be working in areas affected by

artificial intelligence to find jobs elsewhere. There may be non-

inflationary running room for the Fed to do that. In the longer

term, we generally don’t find that cost saving innovations lead to

higher unemployment rates or significantly lower inflation. In the

short-run that may be the consequence, but in the longer term

when policy reacts, the economy ends up at similar

unemployment and inflation levels.

Terry: Those themes that emerge out of this: AI taking jobs or

robotics taking over labor, is that something over time you don’t

seem as being legitimate?

Hatzius: These fears have been around for many years, and what I

think we can say is that so far they haven’t really been borne out. If

we go back to the early 19th century, there were worries about

mechanized spinning machines and the idea that this would put

large numbers of people out of work. In the short run, that

disruption is something that can have a significant impact, but it’s

not the case that technological progress over the longer stretch of

history has led to higher unemployment rates. That is not the case.

My best guess is not that we would end up with much higher

unemployment rates, because in the end people are going to find

something that needs to be done that requires humans and human

labor. My expectation is that it could be tumultuous impact but I

don’t think it is something that will leave us with a higher

unemployment rate.

Terry: Over the past decade, we’ve seen corporate profits

increasingly going to buybacks and dividends over capital

investment. Is there a threshold where from a macro-economic

perspective productivity needs to be in order to drive investment

and capital?

Hatzius: Investment and productivity are linked, and causality goes

in both directions. In recent years, we’ve had relatively low levels

of investment largely for cyclical reasons because there was still a

lot of spare capacity in the economy and capital stock was

underutilized after the great recession. There wasn’t a strong

economic incentive to invest in new capacity. I think that is on the

mend, we have seen some pickup in investment rates. There is a

Listen to the extended version

bigger contribution from business investment to productivity

growth in the next couple of years than there was in, say, 2010 and

2011. In terms of the causality in the other direction, the

opportunity for productivity growth is a big driver of investment,

depending on the sort of discoveries that are being made in

cutting edge sectors. It seems like there are still some significant

discoveries and if that continues to be the case, then there will also

be an incentive to invest.

Terry: When we see gains in productivity historically, how do

those typically impact corporate profits? Do costs simply move to

another part of the income statement as companies seek

competitive advantage or do we actually see sustainable

increases in profitability?

Hatzius: My reading of the historical evidence is that initially a

productivity improvement falls to the bottom line in the company

that has that opportunity, but eventually those high returns get

competed away because more entrants try to get a piece of the

action. It can be sustainable for some period of time, but over the

longer term presumably if the market mechanism is working it will

be competed away.

Terry: To the extent that we see technology driven improvements

and efficiency, what impact do you tend to think that has on asset

valuation? In the 90s, we saw a related market reaction to the

productivity that you were talking about, what is the potential for

something like that to repeat itself to the extent that we see this

kind of productivity improvement around Artificial Intelligence

and machine learning?

Hatzius: As far as the overall economy is concerned, I do think that

if you had evidence of a more sustained productivity

reacceleration and if you found that a lot of the fears that surfaced

in recent years that were stuck with this subpar productivity

growth pace went away, I think you would probably see a

revaluation of equities.

In particular, keeping all else the same, we find that periods of

faster productivity growth also mean periods of higher asset

valuations. If we look at the 1990s, we did see that. We did have a

large bubble that developed towards the end of that period and the

aftermath was quite painful. These things can be temporary, but I

think we do typically see a revaluation.

Heath and Jan continue the conversation in a

video interview.

Watch Now 

via：资料来源网络，起点学院学员收集

起点学院www.qidianla.com，互联网黄埔军校，打造最专业最系统的产品、运营、交互课程。

November 14, 2016 Profiles in Innovation

Goldman Sachs Global Investment Research 20

The Ecosystem: Cloud services, open source key beneficiaries of

the coming investment cycle in AI

We believe the ability to leverage AI technologies will become one of the major defining

attributes of competitive advantage across all major industries in the coming years. While

the strategy will differ by company size and industry, management teams that don’t focus

on leading in AI and benefiting from the resulting product innovation, labor efficiencies,

and capital leverage risk being left behind. Accordingly, we believe the need for

companies to invest in these new technologies to stay competitive will drive a boom in

demand for the talent, services, and hardware underlying artificial intelligence.

As a comparison, the 1990s tech-driven productivity boom drove a corresponding boom in

enablers. Increased capital spending on technology drove an increase in business

formation to capture this capital spending. Software, hardware and networking company

formation inflected, before the inevitable industry consolidation occurred. Exhibit 13 below

highlights this pattern within the software industry. The number of public software

companies with between $200mn and $5bn in inflation adjusted market capitalization

nearly tripled in the 1995-1999 period, before consolidating in the mid-2000s.

Exhibit 13: Rapid growth in the enabler ecosystem

accompanied the 1990s productivity boom

# of software companies with inflation adjusted market cap

$200mn-$5bn (2015 dollars)

Exhibit 14: Venture capital investing in AI has inflected

this decade

Total VC funding in AI/ML by year

Source: Factset, Goldman Sachs Investment Research Source: PitchBook

We see potential for a similar boom related to the coming cycle of AI driven productivity,

with value being created across software, hardware, data, and services providers as

enterprises invest to capitalize on the potential of AI. Reflecting this opportunity, and as

highlighted in Exhibit 14 above, VC funding into AI-related start-ups has inflected sharply

this decade. The potential for a coming boom in enterprise AI investment has also started

to drive consolidation. Cloud platforms in particular have invested heavily in AI talent, with

Google, Amazon, Microsoft and Salesforce making a combined 17 AI-related acquisitions

since 2014 (Exhibit 5).

We also see some benefit in the contextual comparison of where we are in the

development of AI and ML technologies to historical technology cycles. As with other

major technology cycles of the last 50 years, compute (and Moore’s Law) has been both

the inhibitor and the enabler of progress. For example, in systems architecture we have

63 63

64 64

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

$596

$1,589

$3,448

$5,335

$2,786

$4,439

$9,509

$14,586

$12,517

$14,510

$12,883

$13,702

$2,000

$4,000

$6,000

$8,000

$10,000

$12,000

$14,000

$16,000

2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Capital raised in $mn

via：资料来源网络，起点学院学员收集

起点学院www.qidianla.com，互联网黄埔军校，打造最专业最系统的产品、运营、交互课程。

剩余98页未读，继续阅读

黄啊码

粉丝: 1w+
资源: 2313

人工智能技术的发展和应用前景

华为HCIA-AI认证：人工智能概述与发展里程碑

深入理解人工智能学习笔记整理

人工智能：历史、概念与关键技术梳理

2018_Book_ArtificialIntelligence.pdf

life3.0_Being Human in the Age of Artificial Intelligence.pdf

人工智能概述.pdf

先进人工智能研发 2016–2019 PROGRESS REPORT ADVANCING ARTIFICIAL INTELLIGENCE R&D.pdf

第1讲 人工智能与机器学习概述.pdf

[详细完整版]人工智能概述.pdf

人工智能实践：TensorFlow笔记学习（一）——人工智能概述.pdf

最新资源

第1讲人工智能与机器学习概述.pdf