CASE 04
Naming a Driver-Assistance System “Full Self-Driving”
When the disclosure is accurate and the brand name contradicts it
Tesla sells two driver-assistance systems called Autopilot and Full Self-Driving. Both require the driver to watch the road continuously and to be ready to take control at any moment. The owner’s manual says so. The purchase screen says so. The car says so when you switch it on. And then the product is called Full Self-Driving, in every conversation the owner has about the car for the next five years. People have died in crashes where the driver was not watching the road.
Course anchor
Ch. 6 (required) + Ch. 7–18 frameworks
Case period
2015 – present (ongoing)
Decision point
Recurring — keep the name, or change it
The counterweight to Case 02. In the Apple Card case the harm was financial, diffuse, delayed and contested — issue intensity was low and the reaction was enormous. Here the harm is death, immediate, concentrated, and attached to named people whose families testified in public. Running Exhibit 6.7 across both cases is the single most instructive comparison in this case bank, because it shows how badly reaction and intensity can come apart in both directions.
What this case is designed to teach
After analyzing this case, you should be able to:
- Distinguish disclosure from communication (Ch. 14). A company can state something accurately in a document nobody re-reads while its brand name says the opposite every day. Decide which one is the message.
- Apply issue intensity where every factor is at maximum (Exhibit 6.7), and compare with a case where they are not.
- Analyze a control system that can be updated overnight (Ch. 18) — software-defined vehicles allow corrective action at a speed no traditional recall can match, which raises the question of why the correction waited.
- Assess safety claims critically. The company publishes statistics showing its systems are safer. Learn to evaluate what such a comparison does and does not establish, without either accepting or dismissing it.
- Allocate responsibility in a human–machine system where the law places it entirely on the driver and the design invites the driver to relax.
- Evaluate a regulator that publishes findings it cannot enforce — the real-world equivalent of Ch. 6’s independent social audit.
Part A — The Case
Read the whole narrative before looking at the questions. This case involves fatalities; the people concerned were real and are named in public investigation records. Write about them accordingly.
A.1 Background: what these systems actually are
Vehicle automation is conventionally described using a six-level scale published by SAE International and widely used by regulators. The boundary that matters is between Level 2 and Level 3. At Level 2 the system controls steering and speed, but the human is driving: they must monitor the road continuously and remain responsible for everything the vehicle does. At Level 3 and above, the system takes over the driving task within defined conditions and the human may stop monitoring.
Tesla’s Autopilot (introduced 2015) and its Full Self-Driving option (sold from 2016, with a customer “beta” programme from 2020) are Level 2 systems. The driver is driving. Tesla says this clearly in the owner’s manual, in the purchase flow, and in on-screen warnings when the feature is engaged.
Level 2 systems have a well-documented human-factors problem, known long before Tesla existed: people are poor at sustaining attention on a task a machine is performing competently. The better the system works, the less the supervisor supervises — and the moments when supervision matters are precisely the rare ones the system cannot handle. This is not a criticism of any particular company. It is the known failure mode of the entire category, and it is why the design question is not only “how good is the automation?” but “how does the design keep the human engaged?”
The management problem in one sentence
The name of the product is one of the strongest determinants of how much a driver will supervise it — and the name is a marketing decision, made by people who are not safety engineers, measured on things that are not safety.
A.2 The names and what they carry
“Autopilot” is borrowed from aviation, where it denotes a system that flies the aircraft while trained, licensed crew monitor it under strict procedures. The public understanding of the word is looser: an autopilot is the thing that means the pilot is not really flying.
“Full Self-Driving” is not borrowed from anywhere. It is a plain-language English phrase and it means what it says. The word full is doing specific work: it forecloses the possibility that the driving is partial.
The names were not incidental to the business. Full Self-Driving was sold as a paid option at prices reported to have reached roughly US$15,000, later also offered by subscription. Customers were paying, in advance, for a capability the company said would arrive. The company’s chief executive publicly forecast the arrival of full autonomy repeatedly over many years, and those forecasts were repeatedly not met. The name therefore did three jobs at once: it described a product, it sold a promise, and it set an expectation in the mind of the person responsible for watching the road.
A.3 What happened
In May 2016, Joshua Brown was killed in Florida when his Model S, operating on Autopilot, drove beneath a tractor-trailer crossing the highway. It was the first widely reported fatality involving the system. The US National Transportation Safety Board (NTSB) investigated and concluded, among other things, that the way the system was designed had permitted the driver to disengage from the driving task for extended periods.
In March 2018, Walter Huang was killed in Mountain View, California, when his Model X on Autopilot struck a highway barrier. The NTSB’s investigation, reported in 2020, was unusually pointed. It criticised the ineffectiveness of Tesla’s driver-monitoring — which at the time inferred attention largely from torque applied to the steering wheel, a measure a distracted driver can satisfy without looking at the road — and criticised the federal regulator, NHTSA, for not requiring safeguards. The Board noted that Tesla had not formally responded to several safety recommendations it had previously issued.
There were further fatal crashes. In August 2021 NHTSA opened a defect investigation into a pattern of Autopilot-engaged vehicles striking stationary emergency vehicles at roadside incidents, later escalating it. In 2022 the California Department of Motor Vehicles filed complaints alleging that Tesla’s marketing of “Autopilot” and “Full Self-Driving” constituted untrue or misleading advertising.
In December 2023 Tesla recalled approximately two million vehicles in the United States — the great majority of its US fleet — to add controls and alerts intended to keep drivers attentive when Autosteer is engaged. Because the vehicles are software-defined, the recall was delivered as an over-the-air update. NHTSA subsequently opened a query into whether that remedy was adequate. Verify the current status of these proceedings, and of any litigation, before citing them.
A.4 The company’s position, stated fairly
You cannot analyse this case without stating the other side properly. Tesla’s position, as expressed across its public materials and legal filings, rests on several claims that are individually reasonable:
- The disclosures are clear and repeated. The manual, the purchase flow, and the in-car interface all state that the driver must supervise and remains responsible. A customer cannot enable these features without acknowledging this.
- A product name is shorthand, not a specification. No reasonable consumer believes “cruise control” controls their cruise, and aviation autopilots have never meant the pilot may sleep.
- Drivers who defeat the safeguards are responsible for doing so. Some of the crashes involved drivers who were demonstrably not watching the road despite repeated warnings.
- The systems save lives on balance. Tesla publishes figures indicating a lower crash rate per mile with Autopilot engaged than without, and than the US average.
- Delay costs lives too. If the technology eventually reduces road deaths substantially, holding it back until it is perfect has its own body count.
How to handle the safety statistics
The company’s comparison has been criticised on methodological grounds that you should understand rather than simply repeat: Autopilot is used disproportionately on divided highways, which are the safest roads; the vehicles are newer and heavier than the fleet average; and drivers may disengage the system precisely when conditions become difficult, moving the hardest miles into the comparison group. None of this shows the systems are unsafe. It shows the comparison does not establish what it is offered to establish. A strong report says exactly that, and does not pretend to know the answer.
A.5 The regulator that cannot compel
Two US bodies appear in this case and they are not the same. The NTSB investigates transport accidents and publishes findings and recommendations. It has no power to compel anyone to do anything. The NHTSA regulates vehicle safety and can compel recalls, but its authority is built around identifying a defect — a concept developed for mechanical parts that fail, and an awkward fit for a system that performs exactly as designed while the human supervising it does not.
The result is a structure worth studying in its own right: the body with the expertise to say what went wrong cannot require anything, and the body that can require things is working with a legal concept the problem does not match. Ch. 6 discusses independent social audits as a mechanism for holding organizations to their own standards. The NTSB is close to a real-world example — expert, independent, public, and entirely dependent on the organization choosing to respond.
A.6 Where things stand
- Driver monitoring improved. Interior camera-based monitoring was added and the December 2023 recall strengthened attention warnings — changes closely resembling what the NTSB had recommended years earlier. Ask what it cost to arrive there by this route.
- The naming softened. Tesla has moved toward qualifying the product name to indicate that supervision is required. Verify the exact current product name and when it changed — this is directly relevant to your recommendation, because it means the company eventually did some version of what critics asked.
- Litigation and regulatory proceedings continue. Cases have been settled, dismissed, and tried, with outcomes in both directions. Do not cite any specific verdict without checking its current status on appeal.
- The industry problem is unresolved. Multiple manufacturers now sell Level 2 systems with confident names. There is still no binding global standard for what such a system may be called, and the human-factors problem is unchanged.
Part B — Exhibits
You may reproduce these exhibits, but you must extend them with your own analysis.
Exhibit 1 — Timeline
| Date | Event | Why it matters for the analysis |
|---|---|---|
| Oct 2015 | Autopilot introduced via software update. | A safety-critical capability delivered like a consumer software feature (Ch. 7). |
| May 2016 | Joshua Brown killed in Florida while Autopilot engaged. | First widely reported fatality; NTSB later finds the design permitted prolonged disengagement. |
| 2016 → | “Full Self-Driving” sold as a paid option; autonomy repeatedly forecast as imminent. | Revenue is taken for a future capability — and the name enters the owner’s daily vocabulary (Ch. 9, Ch. 14). |
| Mar 2018 | Walter Huang killed in Mountain View while Autopilot engaged. | NTSB (2020) criticises ineffective driver monitoring and the regulator’s inaction. |
| 2017–2020 | NTSB issues safety recommendations on driver monitoring and operational limits; several go unanswered by the company. | The Ch. 6 independent-audit problem: expert findings with no power to compel. |
| Oct 2020 | “FSD Beta” released to customers on public roads. | Beta testing of a safety-critical system by untrained users, among uninvolved third parties who never consented. |
| Aug 2021 | NHTSA opens a defect investigation into Autopilot crashes with stationary emergency vehicles; later escalated. | Regulatory response arrives roughly five years after the first fatality (Ch. 3, Ch. 18). |
| 2022 | California DMV files complaints alleging the marketing was untrue or misleading. | The naming question moves from criticism to a formal legal allegation. |
| Dec 2023 | ~2 million vehicles recalled via over-the-air update to strengthen driver-attention controls. | Feedback control executed in days — which sharpens the question of the seven preceding years. |
| 2024 → | Regulator queries remedy adequacy; naming qualified; litigation continues. | Still developing. Verify before citing. |
Exhibit 2 — Where the product sits, and where the name points
The analytical core of this case. Fill in the right-hand column yourself before you write anything else.
| Level | Who is driving, and what the human must do | Where the product and the name sit |
|---|---|---|
| 0–1 | Human drives. The system may assist with steering or speed, not both. | Conventional cruise control and lane-keeping. |
| 2 | Human drives. System controls steering and speed; human must monitor continuously and is fully responsible. | This is where Autopilot and Full Self-Driving actually are. |
| 3 | System drives within defined conditions; human may stop monitoring but must take over when asked. | — |
| 4 | System drives within defined conditions and can bring itself to a safe stop; no human fallback needed within those conditions. | — |
| 5 | System drives everywhere a human could. No human involvement required at all. | This is what the phrase “full self-driving” means in ordinary English. |
The gap between the row the product occupies and the row its name describes is the entire case. Measure it before you argue about it.
Exhibit 3 — Three messages reaching one driver
| Channel | What it says | How often it is received | How much it shapes behaviour |
|---|---|---|---|
| The legal channel Manual, purchase flow, warnings | You are driving. Watch the road. You are responsible. | Once at purchase; a dismissible prompt thereafter. | Low after the first week. Its function is to allocate liability, not to change conduct. |
| The brand channel The product name | The car drives itself, fully. | Every single time the owner or anyone else refers to the feature. | High. It sets the mental model, and mental models govern attention. |
| The experience channel The system’s own competence | Thousands of uneventful miles. | Continuously. | Highest. Nothing teaches a supervisor to stop supervising faster than a machine that never needs them. |
When these three channels conflict, which one is “what the company communicated”? Your answer to that question will determine most of your report.
Exhibit 4 — Stakeholder inventory (starting point only)
| Stakeholder | What is at stake | Consented to the risk? | Power to affect the outcome |
|---|---|---|---|
| Drivers who bought the system | Their lives; money paid for a promised capability. | Yes — they accepted stated terms. But see Exhibit 3 on what they actually understood. | Low individually; meaningful collectively as purchasers. |
| Passengers | Their lives. | Not meaningfully. | None. |
| Other road users & pedestrians | Their lives. | No. They were never asked. | None. The most important row in this table. |
| Emergency responders | Their lives, at roadside incidents — the subject of a specific regulatory investigation. | No. | None directly; significant through public testimony. |
| Tesla and its shareholders | Revenue from the option, brand differentiation, litigation and regulatory exposure. | n/a | Total control over naming, design, and the timing of updates. |
| Tesla engineers | Professional judgement about what is safe to ship and what to call it. | n/a | High internally; publicly invisible. |
| NTSB | Its own relevance as a safety authority. | n/a | Expertise and publication — no power to compel. |
| NHTSA | Statutory duty for vehicle safety. | n/a | Can compel recalls — but must first establish a “defect.” |
| Other manufacturers | Competitive pressure to match confident claims. | n/a | Collective, through industry norms. |
| Future road users | Whether the technology matures safely, and whether naming norms get set now. | No; not present. | None. |
Exhibit 5 — Key figures
Level 2
The automation level of both products — the human is driving
~2 million
US vehicles recalled in Dec 2023 for driver-attention controls
~7 years
From the first reported fatality to that fleet-wide remedy
~US$15k
Reported peak price of the Full Self-Driving option
Data caution. Crash counts attributed to these systems vary by source and by counting rule (was the system engaged at impact, or in the seconds before?), and figures are revised as investigations conclude. Do not present a total without stating the source, the date, and the counting rule. The defensible numbers here are the structural ones in this exhibit.
Part C — The Assignment
C.1 How this case satisfies the syllabus requirements
| Req. | Syllabus wording | How you meet it in this case |
|---|---|---|
| (1) | Select a real organization or industry case and analyze it using the frameworks covered in this course. | Tesla is a real organization with an extensive public regulatory and investigative record. Apply Ch. 7–18 frameworks in Questions 6–9 — at minimum control (Ch. 18) and communication (Ch. 14). |
| (2) | Include an explicit business ethics and social responsibility analysis: identify the ethical issues, map the stakeholders, assess conduct against ethical criteria and CSR expectations (Ch. 6). | Questions 2, 3, 4 and 5 are mandatory. This case is the bank’s reference example for issue intensity at full strength and for non-consenting stakeholders. |
| (3) | Close with a recommended course of action and the reasoning that supports it. | Question 10 — three parts: the naming decision, the safety governance design, and the industry-level rule. |
| (4) | Written report and oral presentation in English. | Part G provides terminology and presentation language support. |
C.2 First decision: which organization are you analyzing?
(a) Tesla
The firm that named, built, priced and shipped the product. The default choice and the one with the clearest duties.
(b) The US safety regulators
NHTSA and NTSB as organizations with a mission, a mandate, and a gap between them. Excellent for Ch. 11 and Ch. 18; unusual and rewarding.
(c) The driver-assistance industry
Naming and marketing norms across manufacturers. Requires evidence beyond one company — only choose it if you will gather that.
(d) The driver–vehicle system
Treat the human and the machine as one designed system. Conceptually the strongest frame for the ethics, and the hardest to execute.
C.3 Five traps to avoid
- Making this about one executive. This is a case about a naming and safety-governance decision inside a company. A report that becomes a commentary on an individual’s personality has stopped doing management analysis, and will be marked as such.
- Treating the safety statistics as settled — in either direction. The company’s comparison is contested on methodological grounds; that does not make the systems dangerous. Say precisely what the evidence does and does not establish.
- Ignoring driver responsibility. In several crashes the driver was demonstrably not watching the road despite warnings. The law places responsibility on them. A serious analysis engages this rather than assuming the company bears all of it.
- Assuming a name cannot matter because a disclosure exists. That is the position under examination, not a fact you may assume. Exhibit 3 exists to make you argue it.
- Writing carelessly about people who died. The victims are named in public investigation records and have surviving families. Use their names accurately or not at all, describe the crashes factually, and do not speculate about their state of mind beyond what investigators established.
Part D — Analysis Questions
Questions marked REQUIRED must be answered by every team. From those marked CHOOSE, answer at least three, including at least one from the Ch. 14 / Ch. 18 pair.
Context and unit of analysis REQUIRED
Define your unit of analysis (C.2) and describe the external environment. Give particular attention to the legal/political component: a regulatory regime designed around mechanical defects and periodic physical recalls, applied to a product that changes by software overnight and whose failure mode involves human attention rather than component failure. Then characterise the organizational culture using Ch. 3 dimensions — specifically, what happens when a culture that rewards speed, boldness and iteration is applied to a safety-critical product, and whether such a culture can hold both values at once.
Frameworks: Ch. 3 external environment components; environmental uncertainty; dimensions of organizational culture; strong culture and its risks.
Identify and rank the ethical issues REQUIRED
Identify at least four distinct ethical issues, each stated as a decision faced by a named actor. Candidates include: choosing and retaining the product names; selling a capability in advance of its existence; releasing a “beta” of a safety-critical system to untrained users on public roads; relying on a driver-monitoring method known to be defeatable; not formally responding to published safety recommendations; and publishing a safety comparison the company knows is not like-for-like.
Rank them using issue intensity (Exhibit 6.7), factor by factor. You will find this case scores at or near maximum on nearly every determinant — greatness of harm, probability, immediacy, proximity, concentration — with consensus of wrong as the only genuinely contested factor. Then do the comparison that makes this question worth answering: run the same six factors over Case 02 (Apple Card), where the reaction was enormous and the intensity low. Explain what the two cases together reveal about the relationship between issue intensity, public reaction, and how quickly an organization actually changes its behaviour.
Frameworks: Exhibit 6.7 issue intensity; ethical dilemmas; consent and harm to non-consenting parties.
Map the stakeholders REQUIRED
Extend Exhibit 4, recording stake, consent, power, legitimacy and outcome. Then answer three harder questions:
- What is owed to people who never agreed to anything? The driver accepted terms. The pedestrian, the oncoming driver, and the emergency responder at a roadside did not. Ch. 6’s treatment of responsibility is built around obligations beyond the immediate transaction — apply it, and state what duty is owed to a stakeholder who cannot consent, cannot opt out, and gets no benefit.
- Where does responsibility sit between the driver and the designer? The law says the driver. The design predictably produces the behaviour that causes the harm. Argue the allocation explicitly rather than assuming it, and consider whether a system that is legally safe and behaviourally unsafe is acceptable.
- Are the engineers stakeholders? Somebody inside the company had a professional view about the driver-monitoring method and about the name. Locate them, and ask what channel Ch. 6 says they should have had.
Frameworks: Stakeholder identification and prioritization; power–legitimacy–urgency; consent; Ch. 6 CSR expectations; whistle-blower protection.
Assess conduct against ethical criteria and CSR expectations REQUIRED
This question has three parts, all required.
(a) Apply the ethics tests. Take the five tests in Exhibit 6.4 and apply them to three decisions: (i) naming the product “Full Self-Driving”, (ii) releasing the beta to customers on public roads, and (iii) not formally responding to the safety recommendations. Present a 3×5 matrix. The Golden Rule test has an unusually literal form here: would you be content for your own family to be the pedestrians beside a vehicle whose supervising driver believes the car drives itself? And the What-If-Everybody test bites hardest on the naming — work out what the road looks like if every manufacturer names its Level 2 system as ambitiously as the market allows.
(b) Classify the CSR posture. Classify the conduct as social obligation, social responsiveness, or social responsibility, with evidence. Apply Exhibit 6.2 to the December 2023 recall specifically: ethical or pragmatic? ends or means? long-term or short-term? Note the difficulty and address it head-on: the company did eventually make substantially the changes the safety authority had recommended, but did so after years, an escalating investigation, and an advertising complaint. Does a correct action taken under pressure count as social responsibility? Answer using the framework, not your instinct.
(c) Classical vs. socioeconomic view. State the strongest classical-view defence: the company meets every legal requirement, discloses accurately, competes lawfully, and its product may on balance reduce deaths — setting the safety standard is the regulator’s job, not a manufacturer’s. Then give the socioeconomic reply. Finally take a position on the hardest version of the question: if a product plausibly saves more lives than it costs, does it matter that the way it is sold predictably causes some of the deaths it does cause?
Frameworks: Exhibit 6.4 ethics tests; Exhibit 6.2; Exhibit 6.1; classical and socioeconomic views.
Explain the behavior REQUIRED
Use Exhibit 6.5 to explain conduct rather than judge it. Assume, as is almost certainly true, that the people involved wanted the product to be safe.
- Structural variables. Ch. 6 warns that under the pressure of ambitious goals, otherwise ethical employees may do whatever is needed to meet them. Public commitments to imminent autonomy created a goal that could not be revised without an admission. Explain how that produces escalation of commitment (Ch. 2) and what it does to internal dissent.
- Who owns the name? Product naming sits with marketing; driver behaviour sits with safety engineering. Identify the structural gap that allows a decision with safety consequences to be made by a function with no safety accountability — and name the mechanism that should close it.
- Leadership and tone at the top. Ch. 6 states that the behavior of managers is the single most important influence on ethical conduct. Where the most senior figure is also the loudest public advocate of the product’s capabilities, what does that do to an engineer who believes the name is unsafe?
- Issue intensity from the inside. Each individual crash is one event among millions of drives. Explain how an organization can face maximum-intensity harm externally while perceiving it internally as a statistical rate.
Frameworks: Exhibit 6.5 determinants; Exhibit 6.6 moral development; structural variables and goal pressure; Ch. 2 escalation of commitment; leadership at the top.
Control: seven years to a one-day fix CHOOSE
Apply the three types of control (Ch. 18). Feedforward: what should have constrained the product before release — restricting operation to the road types it was designed for, requiring camera-based attention monitoring from the start, a naming review with safety sign-off? Estimate what each would have cost commercially. Concurrent: the company receives telemetry from its entire fleet continuously. It is among the best-instrumented organizations in the world for detecting exactly this problem. What would a well-designed monitoring system have surfaced, and what standard should have triggered action? Feedback: the December 2023 recall reached two million vehicles overnight.
Then the question this case exists to ask: a firm that can correct its product globally in a single day has almost no technical excuse for a seven-year delay. If the barrier was not technical, name what it was — and say what control mechanism would have overcome it. Relate your answer to Ch. 6’s independent social audits, using the NTSB as the real-world instance of an expert auditor with no power to compel.
Frameworks: Ch. 18 control process, feedforward/concurrent/feedback control, corrective action, control criteria; Ch. 6 independent social audits.
Communication: which channel is the message? CHOOSE
Using Ch. 14 and Exhibit 3, analyse the three channels reaching the driver and resolve the conflict between them. Apply the communication process model: who is the sender, what is encoded, through what channel, and what is actually decoded by the receiver? Identify the barriers — including the one that matters most here, which is that a warning acknowledged once is a compliance act, while a name repeated daily is communication.
Then argue the general proposition, since it applies far beyond cars: when an organization’s formal disclosure and its informal signals conflict, which one has it communicated — and which one should it be held responsible for? Test your answer against a case where it cuts the other way, such as a product whose name is more cautious than its actual capability.
Frameworks: Ch. 14 communication process, barriers, nonverbal and symbolic communication; marketing ethics; Ch. 15 perception and mental models.
Disruptive innovation and the cost of speed CHOOSE
Using Ch. 7, analyse the company as a disruptive innovator in an industry with century-old safety conventions. Shipping a “beta” to consumers is normal in software and unprecedented for a safety-critical vehicle system. Assess what transfers from software practice to vehicles and what does not, and be specific about where the boundary falls.
Then take the strategy angle (Ch. 9). The name was a genuine source of competitive differentiation and a direct revenue line. Evaluate it with a SWOT, and assess whether an advantage that depends on a claim the product cannot yet support is sustainable. Finally, consider the industry effect: if confident naming sells, competitors face pressure to match it. Is this a case where the first mover set an industry norm that no single firm could then afford to break — and if so, what does that imply about where the remedy has to come from?
Frameworks: Ch. 7 disruptive innovation, change and resistance; Ch. 9 SWOT, competitive advantage, sustainability; industry-level collective action problems.
Decision making under escalating commitment CHOOSE
Using Ch. 2, classify the naming decision and each subsequent decision to retain the name. Identify the biases operating, with the behavior that evidences each: escalation of commitment to public predictions and to customers who had already paid; confirmation bias in how internal safety data was read; overconfidence in the timeline; and availability in weighing vivid individual crashes against an aggregate rate.
Then design the decision process. Who should have to approve a product name with safety implications? What evidence should be required? What would have made it possible to change the name without it being an admission of liability — a genuine constraint, since the legal cost of conceding is exactly what keeps such names in place across many industries.
Frameworks: Ch. 2 decision types, conditions, biases and errors; escalation of commitment; Ch. 8 goal setting and the risks of public commitments.
Your recommendation REQUIRED
(a) The naming decision. Place your team in the company in 2021 — after the NTSB findings, before the regulator’s escalation. Recommend what to do about the product names. Your recommendation must state: the decision criteria and their weights; who signs it off; what the product is called and what customers are told; what you offer the customers who already paid for “Full Self-Driving” on the basis of the name; what it costs in revenue, litigation exposure and credibility, and who bears that cost; the strongest argument for keeping the name, stated fairly and then answered; and the conditions under which you would revisit. Note the trap: changing the name can be characterised as admitting the old one was misleading, which has legal consequences. A recommendation that does not confront this has not engaged with the real decision.
(b) The safety governance design. Design the internal governance you would require before shipping any capability whose safe use depends on sustained human attention. Address at least: a naming review with safety sign-off and the authority to refuse; pre-release human-factors testing of what users actually do rather than what the manual instructs; continuous fleet monitoring with pre-declared thresholds that trigger action automatically; a formal, published response to every external safety recommendation, whether accepted or rejected, with reasons; and a protected channel for engineers who believe a release is unsafe. For each, name the Ch. 6 mechanism it implements and the specific failure it addresses.
(c) The industry rule. Individual firms face competitive pressure to name ambitiously, so a single firm restraining itself is penalised. Recommend how the naming of driver-assistance systems should be governed — binding regulation, an enforced industry standard, or something else — and say who would have to agree, what it would cost, and why your mechanism would not simply be ignored the way the safety recommendations were.
Frameworks: Ch. 2 decision-making process; Ch. 6 codes of ethics, independent social audits, whistle-blower protection, ethics training; Ch. 18 control design; Ch. 8 planning.
Optional extensions
- Comparison with Case 02. Run Exhibit 6.7 across both and explain why reaction and intensity came apart in opposite directions. The strongest available extension for this case.
- International context (Ch. 4). Regulators in different jurisdictions have taken different positions on what these systems may be called and where they may operate. Apply Ch. 6’s treatment of international ethics: should a firm apply the strictest standard globally, or the local one in each market?
- Green management (LO 6.2). The same company is a major contributor to transport electrification. Assess whether strong performance on one dimension of social responsibility has any bearing on how its conduct on another should be judged — a question that recurs across the whole field.
- Operations (Ch. 18b). Compare over-the-air software recalls with traditional physical recalls as quality-management mechanisms. What does the ability to fix everything overnight do to the incentive to get it right first?
Part E — Deliverables
Written report
- Length: 3,000–4,000 words of body text (roughly 10–14 pages), excluding cover page, exhibits, and references.
- Language: English throughout.
- Format: A4, 12 pt, 1.5 line spacing, numbered pages and numbered section headings.
- Citations: Every factual claim and every number carries a source. Consistent style (APA 7th recommended). A reference list is mandatory.
- Contribution statement: a short table naming which member led which section, signed by all members.
Required structure
- Executive summary (max. 250 words, written last)
- Case background and unit of analysis (Q1)
- Ethics and social responsibility analysis (Q2–Q5) — the largest section
- Management analysis (your three chosen questions from Q6–Q9)
- Recommendation and reasoning (Q10)
- Limitations and what would change our answer
- References · Appendices · Contribution statement
Oral presentation
- Time: 15 minutes presentation + 5 minutes Q&A (confirm in class).
- Participation: every team member speaks for a meaningful segment.
- Slides: maximum 15. Any slide that is only a paragraph of text will cost you marks.
- Language: English throughout, including Q&A.
- Tone: people died in this case. Do not use crash imagery for effect, and do not treat the fatalities as a rhetorical device.
Suggested timing
- 2 min — The gap between the product and the name, and your recommendation up front
- 2 min — SAE levels and the human-factors problem (assume no prior knowledge)
- 6 min — Ethics and stakeholder analysis, including non-consenting parties
- 3 min — Control and communication analysis
- 2 min — Recommendation, costs, strongest counter-argument
Prepare for: If the system saves lives on balance, why should they change anything? Isn’t the driver legally responsible? What would your rule cost, and would it have prevented these specific crashes?
Sourcing and academic integrity
- Prefer primary sources: NTSB accident reports and safety recommendations, NHTSA investigation documents and recall notices, the California DMV filings, the company’s owner’s manuals and published statements, and court documents. These are unusually accessible for this case — use them.
- Crash counts are a trap. Different sources use different counting rules and figures are revised as investigations close. Always state source, date, and counting rule.
- Litigation changes. Verdicts are appealed, settlements are confidential, and investigations close without findings. Check the current status of anything you cite.
- Product names and features change. Confirm the current naming and driver-monitoring implementation before writing about them — this directly affects whether your recommendation has already been adopted.
- If you use generative AI tools, disclose what you used them for in an appendix. You remain fully responsible for every claim. This case is high-risk for invented statistics: a tool will readily produce a confident-sounding crash total that no source supports.
Part F — Grading Rubric
Total 100 points, mapped to the four syllabus requirements. The final group report is worth 30% of the course grade.
| Pts | Criterion | Excellent | Weak |
|---|---|---|---|
| 15 | Case understanding & framing Req. (1) · Q1 |
Unit of analysis declared and held; SAE levels used correctly; the company’s position stated fairly before being assessed; the safety statistics handled with precision. | Becomes a commentary on an individual; treats the systems as autonomous or as useless; quotes crash totals with no source or counting rule. |
| 30 | Ethics & CSR analysis Req. (2) · Q2, Q4, Q5 |
Four or more issues identified and ranked; issue intensity worked factor by factor and compared with a low-intensity case; non-consenting stakeholders taken seriously; the classical defence stated at full strength before being answered. | Moral verdict substituted for analysis; the company’s position dismissed rather than engaged; driver responsibility ignored; fatalities used for rhetorical effect. |
| 15 | Stakeholder mapping Req. (2) · Q3 |
Goes beyond the exhibit; consent used as an analytical dimension; pedestrians, responders and future road users analysed; the driver/designer allocation argued rather than assumed. | Exhibit 4 reproduced; only drivers and the company discussed; consent not distinguished from stake. |
| 15 | Management frameworks Req. (1) · Q6–Q9 |
At least three frameworks from Ch. 7–18 applied to generate insight; the seven-years-to-one-day control question confronted directly. | Frameworks decorate the report; control types listed but not applied; conclusions unchanged if the frameworks were deleted. |
| 15 | Recommendation & reasoning Req. (3) · Q10 |
Specific and actionable; confronts the admission-of-liability trap; addresses customers who already paid; industry-level mechanism proposed with a reason it would not be ignored; costs named and allocated. | “Change the name” with no consideration of cost or legal consequence; no answer for existing purchasers; industry rule asserted with no enforcement mechanism. |
| 10 | English communication Req. (4) |
Clear, well-organized English; automation and safety terminology used correctly; presentation delivered, not read; appropriate tone throughout; Q&A handled with composure. | Meaning obscured by language errors; “autonomous” and “driver assistance” used interchangeably; slides read aloud; insensitive treatment of fatalities. |
Note on fluency: grammatical errors that do not obstruct meaning are not penalized. Misusing a management or technical concept is penalized — in this case, using “autonomous”, “self-driving” and “driver assistance” as synonyms is a content error, because the distinction is what the case is about.
Part G — EMI Toolkit
G.1 Key terms
| Term | Meaning in this case |
|---|---|
| Driver assistance (ADAS) | Systems that help the driver but do not replace them. Levels 0–2. Use this term for the products in this case. |
| Autonomous / self-driving | A vehicle that drives itself without human supervision. Levels 4–5. Not what these products are. Using this word loosely is a content error. |
| SAE levels | The 0–5 scale describing how much of the driving task is automated and what the human must do. See Exhibit 2. |
| Operational design domain (ODD) | The conditions a system was designed for — road types, weather, speeds. Limiting a system to its ODD is a feedforward control. |
| Driver monitoring | How the vehicle checks the human is paying attention. Steering-wheel torque can be satisfied without looking at the road; camera-based gaze monitoring is harder to defeat. |
| Automation complacency | The well-documented tendency of human supervisors to reduce attention as a system proves reliable. The central human-factors problem of Level 2. |
| Human factors | The discipline studying how people actually behave with systems, as opposed to how designers intend them to. |
| Beta | Software released to users before it is considered finished. Ordinary in consumer software; contested for a safety-critical vehicle system on public roads. |
| Over-the-air update | A software change delivered remotely to vehicles already sold. Makes correction extremely fast — which changes what counts as a reasonable delay. |
| Recall | A formal remedy for a safety defect. Note that a recall may now be a software update requiring no physical action from the owner. |
| Defect | The legal trigger for regulatory action. A concept built for parts that fail, which fits awkwardly where the system works as designed and the human does not. |
| Safety recommendation | A formal published proposal from an investigative body. Expert and public — and, crucially, not binding. |
| Informed consent | Agreement to a risk by someone who understands it. Applies at best to the driver; not at all to pedestrians and other road users. |
| Escalation of commitment | Increasing investment in a course of action to justify earlier decisions. Directly relevant to repeated public predictions. |
G.2 Useful language for the presentation
Being precise about the technology
- “These are Level 2 driver-assistance systems. The human is driving.”
- “Our argument is about the gap between the product’s level and its name.”
- “The system performed as designed; the question is what the design invited the driver to do.”
Handling the statistics
- “The comparison is not like-for-like, so it does not establish what it is offered to establish.”
- “We are not claiming the system is more dangerous. We are saying the published evidence does not settle it.”
- “Counting rules differ across sources, so we state ours.”
Conceding, then answering
- “The driver is legally responsible. We accept that. Our question is whether the design made that responsibility realistic.”
- “Delay has a cost in lives too. We take that seriously, and here is why it does not settle the naming question.”
- “The company has since changed this. We are assessing the seven years before it did.”
Avoid these
- “Self-driving cars killed people.” — these are not self-driving cars; the error undermines your whole argument
- “The CEO is reckless.” — a personality claim, not an analysis
- “They should just rename it.” — not a recommendation until you have priced it and dealt with existing purchasers
Part H — Sources and Further Reading
Primary sources (start here)
- NTSB accident reports and safety recommendations on the 2016 Florida and 2018 Mountain View crashes, at ntsb.gov. These are the best sources in the case — independent, expert, detailed, and publicly available.
- NHTSA investigation documents and the December 2023 recall notice (Part 573 report), at nhtsa.gov.
- California DMV filings concerning the marketing of “Autopilot” and “Full Self-Driving.”
- Tesla’s owner’s manuals, in-vehicle disclosures, purchase-flow language, and published vehicle safety reports.
- SAE J3016, for the definition of the automation levels.
Course materials
- Robbins, S. P. & Coulter, M., Management, 15th ed., Global Edition, Pearson, 2021 — Ch. 6 (Exhibits 6.1–6.8), Ch. 2, 3, 7, 9, 14, 15, 18, 18b.
Secondary sources
There is a substantial academic literature on automation complacency and human–automation interaction that predates this case by decades — it is the strongest theoretical foundation available to you. Search your library databases for “automation complacency”, “automation bias”, and “vigilance decrement”, and separately for “driver monitoring” AND “partial automation”. Using this literature is what separates an analysis from an opinion.
A note on the facts in this document. This case summarises the public investigative and regulatory record. Investigations, litigation, product naming and driver-monitoring implementations have all continued to change. Verify anything you rely on, date your statements, and report uncertainty where it exists. Do not cite this case document itself as the source of a fact in your report.