The transformation to zero-emission engines for heavy-duty vehicles is expected to go fastest for city buses. Markus Grabitz analyzed the Commission’s proposal for carbon regulation of heavy-duty vehicles presented yesterday.
The European Court of Auditors wants to keep a close eye on the Commission and member states as they implement the NextGenerationEU construction program. In an interview with Christof Roche, Klaus-Heiner Lehne, a German member of the Court, says where he sees potential need for improvement.
The EU Commission could present its proposals for reforming European debt rules later than previously planned. It was said in Commission circles on Tuesday that the backdrop to this are the profound differences among the member states. So far, the authority wanted to present the package shortly after the EU summit at the end of March. Commission Vice President Valdis Dombrovskis appealed to governments yesterday to reach a broad consensus by the summit. You can read more about this in the news section by Till Hoppe and Christof Roche.
Today, the shadow rapporteurs sit together for five hours to finally reach decisions on the AI Act. The EU’s proposed regulation relies heavily on the idea of humans monitoring AI systems to prevent harmful decisions in high-risk applications. But AI expert Johannes Walter explains in his opinion that this doesn’t work in every case.
The transformation to zero-emission engines for heavy-duty vehicles is expected to go fastest for city buses. Starting in 2030, all new city buses in the EU are to stop emitting carbon. This is provided for in the proposal for the carbon regulation for heavy-duty vehicles, which the Commission adopted yesterday.
The Commission leaves one loophole: If the municipality’s terrain is mountainous and climate conditions are particularly unfavorable for e-buses, the Commission can grant the member state exemptions from the 100 percent zero-emissions rule from 2030.
Buses and light trucks will be subject to carbon regulation starting in 2030. Until now, the regulation has only applied to heavy trucks. The Commission has adopted the following benchmarks for reducing the carbon fleet limit value per manufacturer, based on the reference year 2019:
The Commissioners have agreed on these figures. The Commission Vice-President responsible for the Green Deal, Frans Timmermans, wanted to push through a total phase-out of internal combustion engines for heavy-duty vehicles as well and reportedly failed due to the resistance of Transport Commissioner Adina Valean and Industry Commissioner Thierry Breton. Timmermans had gone into negotiations with his colleagues with minus 50 percent for 2030 and a total ban on internal combustion vehicles by 2040.
According to the Commission’s proposal, a zero-emission vehicle emits less than five grams of carbon per ton kilometer. Accordingly, low-emission vehicles are commercial vehicles that emit less than half the reference value.
This makes it clear that zero-emission vehicles include battery, fuel cell, and hydrogen combustion engine drive systems. At his press conference, Timmermans mentioned the hydrogen combustion engine as a possibility for a carbon-free drive for the first time.
The current proposal is an adjustment to existing legislation, which calls for minus 15 percent in 2025 and minus 30 percent in 2030.
The fleet limits apply individually to each manufacturer. If the manufacturer fails to meet the targets, it faces heavy penalties. Starting in 2035, a fine of €4250 will be due for every gram and ton kilometer a manufacturer fails to meet the fleet limit values. One major German manufacturer calculated that missing the targets by five percent a year would cost it billions.
The industry reacted with shock to the Commission’s proposals. It regards the 2030 target, in particular, as “catastrophic.” It said that fuel-efficient engines and better aerodynamics could achieve savings of just under ten percent in the diesel fleet by 2030. Manufacturers would then have to achieve the remaining 35 percent reduction through zero-emission vehicles.
The industry association ACEA calculates the impact minus 45 percent would have in 2030: 400,000 zero-emission trucks would have to be on the road, and at least 100,000 zero-emission vehicles would have to be newly registered each year. This would require the creation of 50,000 publicly accessible charging stations for trucks within seven years, including 35,000 high-performance stations. “Given that there are currently virtually no charging stations for trucks in the public domain, the challenge is very ambitious,” says Sigrid De Vries of ACEA.
Benjamin Krieger of CLEPA, the EU’s supplier association, takes a similar view: “Raising the targets for 2030 and 2035 is very challenging.” He adds that it was only four years ago that the now outdated 2030 target was set, “even then, it was ambitious.”
The German Engineering Federation VDMA welcomes that the internal combustion engine will not be banned completely: “In the case of trucks, the EU Commission has avoided the mistake it made when regulating passenger car drives. The narrow thinking and ban on internal combustion engines have now given way to a more open space for technology and innovation. We welcome that,” says Hartmut Rauen of the VDMA.
CDU member of parliament Jens Gieseke says: “The fields of application for heavy-duty vehicles are so diverse that a one-sided bet on pure electromobility cannot work. Competition is thus needed to find the smartest and most efficient solutions.”
Michael Bloss (Greens) complains, “With this proposal, the Commission is jeopardizing its own climate target of becoming climate neutral by 2050 at the latest.” Commission President Ursula von der Leyen is thus counteracting the Green Deal, she said: “Without an end date for fossil truck traffic, there will still be climate-damaging carbon emissions on European roads well beyond the middle of the century.”
The European Court of Auditors (ECA) is taking a close look at the implementation of the multi-billion NextGenerationEU (NGEU) reconstruction and resilience program and is not ruling out improvements to the legal framework. The German member of the ECA, Klaus-Heiner Lehne, said Table.Media that the Court of Auditors will present a report on the review of the European Commission’s control system as early as next month.
“The Court of Auditors is looking at the way the Commission checks whether funds are spent properly in a member state.” There is a need for discussion with the Commission on that at the moment, he said. “In the Commission’s view, after the transfer of the funds the rest seems to be more or less up to the national governments,” he said. According to Lehne, “it is particularly important to insure that control systems work in the member states.”
He stressed that the Court will also not rely solely on Commission monitoring for NGEU implementation. “We will also look at national control systems in interaction with NGEU to protect the EU’s financial interests.” But that will take some time, he said. The corresponding report should be published in March 2024, he said. Lehne stressed that the possibilities for the Court of Auditors to look at national control systems are “very extensive.”
Another report, to be published later this year, will look at monitoring the impact of NGEU funds, according to Lehne. An initial Court opinion on the National Recovery and Resilience Plans (ARPs) had raised concerns on this point. “We need clear review mechanisms for milestones and targets using appropriate definitions,” Lehne said.
In addition, the GAO will include NGEU implementation in its usual compliance checks as part of its annual reports. “Although this has not yet been finalized, the compliance checks will focus on disbursements. Regarding targets, such as digital and green transformation, we will take a look at this in special reports,” Lehne said.
He referred to the 2022 annual report, in the course of which the Court of Auditors reviewed the first and until then only disbursement in 2021, which had gone to Spain. He said it was found that the disbursement had been made even though a milestone – a requirement for corporate tax law – had not yet been fully met. “But that was marginal and thus acceptable to us.”
Lehne again stressed that the Commission needs to set quantifiable milestones and targets for member states. “The Commission must clearly indicate the importance and value of each individual reform and investment target for the Member State, so that it is clear what the consequence will be if they fail to meet the targets, for example, with the retrieval of disbursed funds. This is still lacking, and where we believe the Commission now needs to make a definition.”
In this context, Lehne also argued in principle for greater standardization of milestones involving policy reforms. He acknowledged, however, that this would not be an easy exercise due to the ARP’s national focus. “Environmentally, for example, a measure may make sense in one state but not in another.”
However, the lack of certification is an issue, particularly in the area of political reforms for climate and environmental protection, he said, as the Commission plans to finance 30 percent of the NGEU funds of more than €800 billion through the issuance of so-called green bonds. “We will investigate whether the green bonds that the Commission is raising in the capital market for NGEU, which are linked to climate change reforms, may be risky and whether more clarification is needed here.”
According to Lehne, the Court of Auditors also sees potential problems with overlapping financing of projects for which funds from the EU’s structural and investment funds are earmarked at the same time. “Precisely because of the high speed at which enormous sums are flowing into the countries, we must ensure that coordination with the other EU funding sources works and that irregularities and fraud are also prevented. There is still considerable work to be done here.”
In this context, the former ECA president pointed out that the far-reaching audit rights of the Court were explicitly enshrined in the NGEU Regulation. He said the ECA has the possibility to look at whether the objectives have been achieved during implementation up to the final beneficiary. “We are able to have every Member State provide us with every file in detail on every payment transaction,” Lehne said, underlining the Court’s determination to detect systemic sources of error. “We did the same with the Juncker plan.”
In this context, Lehne emphasized good and close cooperation with the national and regional audit offices. In addition, the ECA will recruit 29 new auditors with a view to NGEU control. “We had requested 40 new posts from the Budgetary Control Committee of the EP and Council, but we are very well equipped to fulfill our audit mandate for NGEU with the new staff and through internal redeployment.”
Lehne again stressed that the NGEU approach, with its milestones and targets, “may be the model of the European future if it works and shows success.” He said NGEU is less complex than the traditional budget approach and thus has a more immediate and rapid impact. As a result, he said, all stakeholders, but especially the Commission, “have a massive political interest in making this a success.” The Commission and the Court of Auditors “are pulling in the same direction here,” the former ECA President emphasized.
Nijeer Parks was arrested for a crime he didn’t commit. He was arrested due to an incorrect match produced by a facial recognition algorithm. Being already the third known case of a Black man being wrongfully arrested in the US, these false arrests illustrate the lower accuracy of facial recognition algorithm for Black faces.
But what failed Nijeer Parks was not just technology. Had police officers double-checked the matching images, they would have noticed the suspect and Nijeer Parks did not look alike. The suspect in the photo even wore earrings – whereas Nijeer Parks had no piercings. The human police officers should have overseen the algorithmically generated arrest recommendation but failed to do so.
The EU’s proposed Artificial Intelligence Act (AI Act) bets heavily on the idea of humans overseeing AI decision support systems to prevent harmful outcomes in high-risk applications. Machine learning algorithms increasingly often support human decision-making in settings that are critical for our society: For example, in healthcare, when algorithms recommend which patients should undergo expensive treatments, in hiring decisions, when they suggest which applicant to invite for an interview or in financial loan decisions. In all of these cases a human has the final say and could adjust or even overrule an algorithmically derived recommendation.
Human oversight of AI systems can work well. One study in a child welfare decision-making context in the United Kingdom found that humans were indeed able to identify poor algorithmic advice. For another illustration, consider Large Language Models, like ChatGPT. Presumably many of the readers of this opinion piece have been trying out such models themselves in the last couple of weeks and probably soon encountered situations in which the chat bots produced obviously nonsensical answers. With these examples in mind, it is easy to imagine applications where human oversight has its place.
Yet all too often human oversight fails, like in the case of Nijeer Park. The evidence that humans are often not able to be good supervisors for AI has mounted in recent years. In a new experiment we show that people cannot accurately assess the advice quality of an advising algorithm. Participants of the experiment had to solve a simple task and received advice from a supporting algorithm.
Unbeknownst to them, we had tampered with the algorithm, resulting in poor recommendations. Despite this, our participants remained steadfast in their reliance on the algorithm, failing to recognize the magnitude of its inaccuracy even after multiple rounds of playing the game. This result from the lab is supported by various field studies: Judges, physicians and the police have all been found to be poor algorithm monitors.
The reasons for this finding are as intriguing as they are plentiful. Situations where humans make decisions with the help of algorithms are teeming with psychological effects. For example, to a certain degree, people feel that relying on AI absolves them of their responsibility. Its recommendation sets a seemingly objective default from which it is difficult to deflect.
Often times, when the task is perceived as more abstract and mathematical in nature, humans have been found to rely too much on algorithms, a phenomenon termed “algorithmic appreciation.” In other cases, where the setting is perceived as more subjective, one finds “algorithm aversion,” i.e. people erroneously do not follow superior algorithmic advice.
The insight that humans are not infallible overseers of AI has yet to make its way into the draft for the AI Act. Based on our research, we therefore make three recommendations.
Nijeer Parks was released after ten days in jail and spent around $5.000 for his defense. Instituting more carefully implemented human oversight has the potential to avoid similar injustices in the future.
Johannes Walter is a researcher at the Leibniz-Centre for European Economic Research and is doing his PhD in Technical Economics at KIT. His research focuses on the conditions for the safe use of human-AI decision-making. The research paper on which this opinion piece is based was largely written during a research stay at MIT. Read more in the current policy brief from the Leibniz-Centre for European Economic Research.
The transformation to zero-emission engines for heavy-duty vehicles is expected to go fastest for city buses. Markus Grabitz analyzed the Commission’s proposal for carbon regulation of heavy-duty vehicles presented yesterday.
The European Court of Auditors wants to keep a close eye on the Commission and member states as they implement the NextGenerationEU construction program. In an interview with Christof Roche, Klaus-Heiner Lehne, a German member of the Court, says where he sees potential need for improvement.
The EU Commission could present its proposals for reforming European debt rules later than previously planned. It was said in Commission circles on Tuesday that the backdrop to this are the profound differences among the member states. So far, the authority wanted to present the package shortly after the EU summit at the end of March. Commission Vice President Valdis Dombrovskis appealed to governments yesterday to reach a broad consensus by the summit. You can read more about this in the news section by Till Hoppe and Christof Roche.
Today, the shadow rapporteurs sit together for five hours to finally reach decisions on the AI Act. The EU’s proposed regulation relies heavily on the idea of humans monitoring AI systems to prevent harmful decisions in high-risk applications. But AI expert Johannes Walter explains in his opinion that this doesn’t work in every case.
The transformation to zero-emission engines for heavy-duty vehicles is expected to go fastest for city buses. Starting in 2030, all new city buses in the EU are to stop emitting carbon. This is provided for in the proposal for the carbon regulation for heavy-duty vehicles, which the Commission adopted yesterday.
The Commission leaves one loophole: If the municipality’s terrain is mountainous and climate conditions are particularly unfavorable for e-buses, the Commission can grant the member state exemptions from the 100 percent zero-emissions rule from 2030.
Buses and light trucks will be subject to carbon regulation starting in 2030. Until now, the regulation has only applied to heavy trucks. The Commission has adopted the following benchmarks for reducing the carbon fleet limit value per manufacturer, based on the reference year 2019:
The Commissioners have agreed on these figures. The Commission Vice-President responsible for the Green Deal, Frans Timmermans, wanted to push through a total phase-out of internal combustion engines for heavy-duty vehicles as well and reportedly failed due to the resistance of Transport Commissioner Adina Valean and Industry Commissioner Thierry Breton. Timmermans had gone into negotiations with his colleagues with minus 50 percent for 2030 and a total ban on internal combustion vehicles by 2040.
According to the Commission’s proposal, a zero-emission vehicle emits less than five grams of carbon per ton kilometer. Accordingly, low-emission vehicles are commercial vehicles that emit less than half the reference value.
This makes it clear that zero-emission vehicles include battery, fuel cell, and hydrogen combustion engine drive systems. At his press conference, Timmermans mentioned the hydrogen combustion engine as a possibility for a carbon-free drive for the first time.
The current proposal is an adjustment to existing legislation, which calls for minus 15 percent in 2025 and minus 30 percent in 2030.
The fleet limits apply individually to each manufacturer. If the manufacturer fails to meet the targets, it faces heavy penalties. Starting in 2035, a fine of €4250 will be due for every gram and ton kilometer a manufacturer fails to meet the fleet limit values. One major German manufacturer calculated that missing the targets by five percent a year would cost it billions.
The industry reacted with shock to the Commission’s proposals. It regards the 2030 target, in particular, as “catastrophic.” It said that fuel-efficient engines and better aerodynamics could achieve savings of just under ten percent in the diesel fleet by 2030. Manufacturers would then have to achieve the remaining 35 percent reduction through zero-emission vehicles.
The industry association ACEA calculates the impact minus 45 percent would have in 2030: 400,000 zero-emission trucks would have to be on the road, and at least 100,000 zero-emission vehicles would have to be newly registered each year. This would require the creation of 50,000 publicly accessible charging stations for trucks within seven years, including 35,000 high-performance stations. “Given that there are currently virtually no charging stations for trucks in the public domain, the challenge is very ambitious,” says Sigrid De Vries of ACEA.
Benjamin Krieger of CLEPA, the EU’s supplier association, takes a similar view: “Raising the targets for 2030 and 2035 is very challenging.” He adds that it was only four years ago that the now outdated 2030 target was set, “even then, it was ambitious.”
The German Engineering Federation VDMA welcomes that the internal combustion engine will not be banned completely: “In the case of trucks, the EU Commission has avoided the mistake it made when regulating passenger car drives. The narrow thinking and ban on internal combustion engines have now given way to a more open space for technology and innovation. We welcome that,” says Hartmut Rauen of the VDMA.
CDU member of parliament Jens Gieseke says: “The fields of application for heavy-duty vehicles are so diverse that a one-sided bet on pure electromobility cannot work. Competition is thus needed to find the smartest and most efficient solutions.”
Michael Bloss (Greens) complains, “With this proposal, the Commission is jeopardizing its own climate target of becoming climate neutral by 2050 at the latest.” Commission President Ursula von der Leyen is thus counteracting the Green Deal, she said: “Without an end date for fossil truck traffic, there will still be climate-damaging carbon emissions on European roads well beyond the middle of the century.”
The European Court of Auditors (ECA) is taking a close look at the implementation of the multi-billion NextGenerationEU (NGEU) reconstruction and resilience program and is not ruling out improvements to the legal framework. The German member of the ECA, Klaus-Heiner Lehne, said Table.Media that the Court of Auditors will present a report on the review of the European Commission’s control system as early as next month.
“The Court of Auditors is looking at the way the Commission checks whether funds are spent properly in a member state.” There is a need for discussion with the Commission on that at the moment, he said. “In the Commission’s view, after the transfer of the funds the rest seems to be more or less up to the national governments,” he said. According to Lehne, “it is particularly important to insure that control systems work in the member states.”
He stressed that the Court will also not rely solely on Commission monitoring for NGEU implementation. “We will also look at national control systems in interaction with NGEU to protect the EU’s financial interests.” But that will take some time, he said. The corresponding report should be published in March 2024, he said. Lehne stressed that the possibilities for the Court of Auditors to look at national control systems are “very extensive.”
Another report, to be published later this year, will look at monitoring the impact of NGEU funds, according to Lehne. An initial Court opinion on the National Recovery and Resilience Plans (ARPs) had raised concerns on this point. “We need clear review mechanisms for milestones and targets using appropriate definitions,” Lehne said.
In addition, the GAO will include NGEU implementation in its usual compliance checks as part of its annual reports. “Although this has not yet been finalized, the compliance checks will focus on disbursements. Regarding targets, such as digital and green transformation, we will take a look at this in special reports,” Lehne said.
He referred to the 2022 annual report, in the course of which the Court of Auditors reviewed the first and until then only disbursement in 2021, which had gone to Spain. He said it was found that the disbursement had been made even though a milestone – a requirement for corporate tax law – had not yet been fully met. “But that was marginal and thus acceptable to us.”
Lehne again stressed that the Commission needs to set quantifiable milestones and targets for member states. “The Commission must clearly indicate the importance and value of each individual reform and investment target for the Member State, so that it is clear what the consequence will be if they fail to meet the targets, for example, with the retrieval of disbursed funds. This is still lacking, and where we believe the Commission now needs to make a definition.”
In this context, Lehne also argued in principle for greater standardization of milestones involving policy reforms. He acknowledged, however, that this would not be an easy exercise due to the ARP’s national focus. “Environmentally, for example, a measure may make sense in one state but not in another.”
However, the lack of certification is an issue, particularly in the area of political reforms for climate and environmental protection, he said, as the Commission plans to finance 30 percent of the NGEU funds of more than €800 billion through the issuance of so-called green bonds. “We will investigate whether the green bonds that the Commission is raising in the capital market for NGEU, which are linked to climate change reforms, may be risky and whether more clarification is needed here.”
According to Lehne, the Court of Auditors also sees potential problems with overlapping financing of projects for which funds from the EU’s structural and investment funds are earmarked at the same time. “Precisely because of the high speed at which enormous sums are flowing into the countries, we must ensure that coordination with the other EU funding sources works and that irregularities and fraud are also prevented. There is still considerable work to be done here.”
In this context, the former ECA president pointed out that the far-reaching audit rights of the Court were explicitly enshrined in the NGEU Regulation. He said the ECA has the possibility to look at whether the objectives have been achieved during implementation up to the final beneficiary. “We are able to have every Member State provide us with every file in detail on every payment transaction,” Lehne said, underlining the Court’s determination to detect systemic sources of error. “We did the same with the Juncker plan.”
In this context, Lehne emphasized good and close cooperation with the national and regional audit offices. In addition, the ECA will recruit 29 new auditors with a view to NGEU control. “We had requested 40 new posts from the Budgetary Control Committee of the EP and Council, but we are very well equipped to fulfill our audit mandate for NGEU with the new staff and through internal redeployment.”
Lehne again stressed that the NGEU approach, with its milestones and targets, “may be the model of the European future if it works and shows success.” He said NGEU is less complex than the traditional budget approach and thus has a more immediate and rapid impact. As a result, he said, all stakeholders, but especially the Commission, “have a massive political interest in making this a success.” The Commission and the Court of Auditors “are pulling in the same direction here,” the former ECA President emphasized.
Nijeer Parks was arrested for a crime he didn’t commit. He was arrested due to an incorrect match produced by a facial recognition algorithm. Being already the third known case of a Black man being wrongfully arrested in the US, these false arrests illustrate the lower accuracy of facial recognition algorithm for Black faces.
But what failed Nijeer Parks was not just technology. Had police officers double-checked the matching images, they would have noticed the suspect and Nijeer Parks did not look alike. The suspect in the photo even wore earrings – whereas Nijeer Parks had no piercings. The human police officers should have overseen the algorithmically generated arrest recommendation but failed to do so.
The EU’s proposed Artificial Intelligence Act (AI Act) bets heavily on the idea of humans overseeing AI decision support systems to prevent harmful outcomes in high-risk applications. Machine learning algorithms increasingly often support human decision-making in settings that are critical for our society: For example, in healthcare, when algorithms recommend which patients should undergo expensive treatments, in hiring decisions, when they suggest which applicant to invite for an interview or in financial loan decisions. In all of these cases a human has the final say and could adjust or even overrule an algorithmically derived recommendation.
Human oversight of AI systems can work well. One study in a child welfare decision-making context in the United Kingdom found that humans were indeed able to identify poor algorithmic advice. For another illustration, consider Large Language Models, like ChatGPT. Presumably many of the readers of this opinion piece have been trying out such models themselves in the last couple of weeks and probably soon encountered situations in which the chat bots produced obviously nonsensical answers. With these examples in mind, it is easy to imagine applications where human oversight has its place.
Yet all too often human oversight fails, like in the case of Nijeer Park. The evidence that humans are often not able to be good supervisors for AI has mounted in recent years. In a new experiment we show that people cannot accurately assess the advice quality of an advising algorithm. Participants of the experiment had to solve a simple task and received advice from a supporting algorithm.
Unbeknownst to them, we had tampered with the algorithm, resulting in poor recommendations. Despite this, our participants remained steadfast in their reliance on the algorithm, failing to recognize the magnitude of its inaccuracy even after multiple rounds of playing the game. This result from the lab is supported by various field studies: Judges, physicians and the police have all been found to be poor algorithm monitors.
The reasons for this finding are as intriguing as they are plentiful. Situations where humans make decisions with the help of algorithms are teeming with psychological effects. For example, to a certain degree, people feel that relying on AI absolves them of their responsibility. Its recommendation sets a seemingly objective default from which it is difficult to deflect.
Often times, when the task is perceived as more abstract and mathematical in nature, humans have been found to rely too much on algorithms, a phenomenon termed “algorithmic appreciation.” In other cases, where the setting is perceived as more subjective, one finds “algorithm aversion,” i.e. people erroneously do not follow superior algorithmic advice.
The insight that humans are not infallible overseers of AI has yet to make its way into the draft for the AI Act. Based on our research, we therefore make three recommendations.
Nijeer Parks was released after ten days in jail and spent around $5.000 for his defense. Instituting more carefully implemented human oversight has the potential to avoid similar injustices in the future.
Johannes Walter is a researcher at the Leibniz-Centre for European Economic Research and is doing his PhD in Technical Economics at KIT. His research focuses on the conditions for the safe use of human-AI decision-making. The research paper on which this opinion piece is based was largely written during a research stay at MIT. Read more in the current policy brief from the Leibniz-Centre for European Economic Research.
