1

Expectations on and of our graduates

Eric Grimson, Dept. Head, EECSDuane Boning, Assoc. Dept. Head, EECS

Massachusetts Institute of Technology

Based in part on CRA (Computing Research Association) presentation, summer 2008

2

Hypothesis: Expectations on our graduates are changing

• What is our product?– Our students– We do many other things: new

knowledge, new methods, new ideas, new devices & artifacts…

… but our multiplier is our students• They are the fuel for the

innovation engine

• What are the characteristics of a good product?– A function of the market– The market: where is the

demand for our students?

3

Where are the jobs in general?

• Dynamic growth in science & engineering– Engineering & IT growth far outstrips life science growth

Job Growth (Millions)

0.0

1.0

2.0

3.0

4.0

5.0

6.0

1950 1960 1970 1980 1990 2000

All S&T employees

S&E

Life scientists

Physical scientists

Engineers

Mathematicians/ informationtechnologistsSocial scientists

Technicians

Source: BLS

4

Where are the jobs in general?

• Future projections?

10-year total growth (%)

All occupations

All S&E

Computer/math scientists

Engineers

Life scientists

Physical scientists

Social scientists

S&E managers

Postsecondary teachers

Computer programmers

Healthcare practitioner

0.0 10.0 20.0 30.0 40.0Source: BLS

Conclusion: Large growth projected into the future.*Conclusion: Large growth projected into the future.*

– Engineering growth at ≈ rate of all occupations– Computer related at almost 3x general growth rate

* Projections prior to current

recession

5

Does industry value our graduates?

• Comparative bachelor’s degree starting salary data (national) clearly indicates that they do…

13

24

6

MIT 2008 Average UG & Masters Starting Salaries

7

Industry needs our students, but do our students need industry?

• Where do our undergrads go?– NSF data too

generic– Sampling of MIT

data• Substantial fraction

of undergraduates take non-EECS jobs

• “Other” is primarily finance and consulting

Conclusion: Range of industrial positions is broader than traditional EE & CS industry of 20 years ago

Conclusion: Range of industrial positions is broader than traditional EE & CS industry of 20 years ago

Undergraduate careers

0

5

10

15

20

25

30

35

40

Grad school Large firms Start ups Other Service

Perc

en

tag

e

8

Is this true for PhD graduates as well? Where do they go?

• Harder to find data– NRC data on MIT PhD

grads for past 5 years– Most students head to

traditional industry sectors • List of major employers

are what you would expect -- Google, Microsoft, IBM, Sun, Intel, Analog Devices, TI

Kinds of industry jobs

0

10

20

30

40

50

60

70

80

90

100

Larg

e IT

Small IT

Star

tup

Fina

nce

Consu

lting

Lega

l

Nu

mb

er

9

MIT 2008 Average PhD Starting Salaries

10

Academic vs. Industry Path for PhDs?

– CRA Taulbee survey data for CE/CS PhD students– Total output steady 94-99, slow decline to 02, then upturn,

accelerating around 05• Note that by 07, industry hires outnumber academic hires at PhD schools 2:1

PhD Graduate Destination

0

200

400

600

800

1000

1200

1400

1600

1994 1996 1998 2000 2002 2004 2006

Year

Nu

mb

er

of

Ph

Ds PhD Inst

Non-PhD Inst

Non CS Acad

Industry

Government

Self-Empl

Outside NA

Other

Total

Source: CRA Taulbee, year of graduation

11

Who hires CE/CS PhD students(by numbers)?

– If we separate post-docs from faculty positions, trend is more dramatic• Ratio of industry to tenure track faculty positions is then greater than 3:1• Note major growth in industry hires since 04

PhD production

0

100

200

300

400

500

600

700

800

900

1996 1998 2000 2002 2004 2006

Year

Nu

mb

er

PhD Inst faculty

Post Doc

Non-PhD Inst

Non CS Acad

Industry

Government

Self-Empl

Outside NA

Other

Source: CRA Taulbee, year of graduation

Conclusion: Industry is the major employer of PhD graduates as

well as BS/MS graduates

Conclusion: Industry is the major employer of PhD graduates as

well as BS/MS graduates

12

Challenge: Preparing students for a diverse market & varied career paths

• Multiple options/paths for students– Graduate school

• Science & technology (PhD, Masters)• Law, medical school

– Industry• Large & small; consulting

– Academic options as well

• Key questions: – Are we preparing students well for the market?

• Current needs– Are we preparing students well for their whole

career?• Long term needs & skills

13

So what does industry need?

• Transferable skills– Communications

• oral and written

– Analytic problem solving– Ability to work in a team– Leadership– Use of abstraction and

modularity– Best practices

• Documentation• Testing

14

Importance of skills in career

1 2 3 4 5

New skills on own

Formulate new ideas

Synthesize ideas

Communicate oral

Work on team

Write effectively

Self confidence

In depth knowledge

Lead groups

Quantitative skills

MostImportant

What do our graduates find they actually need?

• Alumni survey from MIT– Classes of 1983,

1988, 1993, 1998, 2003

– Surveyed in 2005– Rated importance of

skills in their career since graduation

– Note where “in depth knowledge” and “quantitative” fall

15

How well did you acquire

0 1 2 3 4 5

New skills on own

Formulate new ideas

Synthesize ideas

Communicate oral

Work on team

Write effectively

Self confidence

In depth knowledge

Lead groups

Quantitative skills

How much do we contribute to our students growth?

• Alumni survey from MIT– Classes of 1983, 1988,

1993, 1998, 2003– Surveyed in 2005– Rated how well their

education experience contributed to growth in skills

– Have worked on communication with new requirements

– Leadership and teamwork issues still need attention

16

So what else does industry need?

• Technical skills– Design (circuits, systems,

software)– Signal processing– Database systems– Systems analysis– Manufacturing technology– Interactive digital media– Human-computer interfaces– …

Conclusion: There is a wide range of industrial needs and a wide range of technical skills -- too

much to expect of any single student

Conclusion: There is a wide range of industrial needs and a wide range of technical skills -- too

much to expect of any single student

17

Hypothesis: Not possible, or even preferable, to teach “everything”

• Too much material to stuff into a four year curriculum– A lot is expected in knowledge and

experience even in traditional areas– Problem is exacerbated when you

factor in need to include experience in related fields depending on area of application or interest

• Computational biology• Environment and energy issues• Social networks• Interactive media• Finance

Conclusion: We need to consider new models of

curricular content and delivery

Conclusion: We need to consider new models of

curricular content and delivery

18

Some possible options for handling explosion of knowledge

• Move to a professional degree– MEng as first professional

degree• Maintain current curricular

structure– But change examples and

scenarios for different student groups

• Change curricular structure– Allow student choice– Tradeoff of some areas with

ancillary areas

19

Move to a professional degree

• 5 Year MEng program– Greater breadth and depth– Capstone experience in large

scale project

– Additional cost burden for students

– Not the right path for every student, so still need 4-year bachelor’s option

20

Preserve the current curriculum but change the examples

• Keep the core sub-disciplines in curriculum– Create variations in each

subarea specialized to student interest

• Algorithms based on biological examples, or on information management, or …

• Systems/control analysis focused on finance, or robotics, or …

• Machine learning applied to biology, or to information management, or …

• Distributed systems for environmental sensing, for information management, for …

– Difficult/costly to field multiple class variants

21

Move to new kinds of degrees

• Acknowledge that not every student can or wants to know everything– Flexibility within degree options

• Provide options across major areas• Allow students some choice

– New degree options• Create specific degrees for new

and emerging areas– Bioelectrical engineering– Computational science and

engineering– Information science

– Enable any (nearly) combination of degrees: create easier path to joint/dual degrees

22

An example: MIT

– 2 introductory subjects– Select 4 of 7 foundation subjects

• 3 specific for CS, 3 specific for EE, 4 of 7 for EECS

– Select 3 header subjects, followed by 2 advanced senior level subjects

• Depth structure enforced• Choices largely based on idea of streams

– Communications, devices, circuits, bioelectrical, software engineering, security, information sciences, HCI, learning, systems, networks, …

– Exploring idea of new degrees• Computational biology

– Replace one of 3 streams in CS degree with a biology stream

23

An example: Georgia Tech CS

• Threads™ (specific paths through curriculum) • Modeling & simulation• Devices• Theory• Information Internetworks• Intelligence• Media• People• Platforms

• Roles (fine tuning of threads based on desired goals of student)• Master practitioner• Entrepreneur• Innovator• Communicator• Policy maker

• Additional degree programs in Interactive Computing and in Computational Science and Engineering

24

What about the expectations of our students?

• Current students have much broader interests than their predecessors– Societal impact– Life science applications– Games and other interactive media– Social computing

• They may not be interested in or need all of the traditional areas of EE and/or CS

• We need to adapt to these needs• We may also benefit by an increased

interest in the field and an increasingly diversified student body

Suggestion: We should pay attention to changing interests and needs of our students

Suggestion: We should pay attention to changing interests and needs of our students

25

Adding elements to existing curricula to meet emerging needs

• As industry changes, do the requirements on curriculum need to change to meet those needs?– Low power devices, low

power computation– Multi-core– Cloud computing– User interfaces for mobile

device

• One approach: keep adding more subjects to cover new areas and technologies

26

Challenges to academia

• Balance teaching fundamentals with needs of specific fields

• Balance teaching foundations of field with changing interests of students

• Ensure that EE/ECE/CS is more than a service to related fields– Contribute to modes of

thought of other fields -- biology, medicine, social sciences, interactive media

27

Conclusions

• Industry remains the primary career path for our graduates (Bachelors, Masters, and PhDs)

• Industry needs our students

• Industry is changing

• Our students are changing too

• We need to adapt to these changes• Must deal with an explosion of new

technologies and areas

• Expanded set of skills are crucial