20
Examination of witnesses
Richard Kempshall, Ben Stephenson and Mark Kenny.
Q27 The Chair: Welcome to this session of the Public Services Committee and our inquiry into police transcriptions. I am very pleased to welcome our three witnesses today from the police force. I am going to ask each of them to introduce themselves and say who they are.
Richard Kempshall: Good afternoon to you. I am a detective inspector on the specialist crime homicide command at the Metropolitan Police. I am also an interview adviser and one of the current force leads in investigative interviewing.
Mark Kenny: Good afternoon, everyone. I am a superintendent with Greater Manchester Police with responsibility for criminal justice.
Ben Stephenson: Good afternoon, committee. I am a detective sergeant in the Metropolitan Police Service. My specific role is in Met prosecutions. I work at Petty France in the Ministry of Justice in a partnership and prosecution reform team. A lot of my work revolves around trying to drive efficiency in the transcription process in the Metropolitan Police.
The Chair: That will be very helpful, I think. Thank you for coming. We look forward to hearing your evidence.
Q28 Lord Blencathra: I have no official interest to declare. Gentlemen, why are suspect interview transcripts read aloud in courts instead of being played, even in those courts where we understand there may be perfect or very good audio and video recordings? What are the risks of that? Are you aware of any cases in your experience where the reading out of transcripts by lawyers gave a different impression from the evidence from the taped interviews? I should say that I calculated that there are about 320 staff in the England and Wales police forces doing transcripts. You in the Met have 89.
Mark Kenny: In terms of evidential presentation, there are probably historic reasons for why transcripts are read within the court. I should say that there is a process that sits before that. You are probably aware that the disclosure of evidence starts first. The police will produce a fully typed transcript of an interview. That will then go to the CPS. Then there is a disclosure process with the defence.
The actual document that goes before the court will be the final agreed document, and that is the one that is read out within the courtroom. That is important to understand, because it is not necessarily the entire transcript. It is what is decided. What will be read will be decided, in effect, by the legal parties as the most relevant parts that pertain to the evidence that they want to cover, and not the whole transcript.
As to why it is read, there is probably a traditional element to that. Recordings are definitely more recent and of course interviews predate that. There will be a key aspect of efficiency within the court scheduling as well. The parties will decide within the court what they want to focus on and really drill down into that area. That is not the police; that is in the hands of the prosecution and the defence. For efficiency and timewise within the court, they will probably go to a very specific area of the evidence within the transcript. In other words, they may not have time to play a full recording, or recordings, which could cover many hours. They may wish to focus on only 20 or 30 minutes within a much longer span of time for the interview.
In terms of risk, personally I would say that there is no system that is perfect. There are risks within any system, and we acknowledge that. For the production of evidence, there are checks and balances within that system prior to it going to the court. The police will present or type the transcript, but then it will go to the defence. They have an opportunity to challenge if there are errors within that transcript. There is a process prior to it being presented within the court.
As to the presentation of it, there is a risk. Traditionally, there is a question and answer process within the trial. Again, that is really for the judge to decide, not for the police to decide. It is for the court to decide how it is presented. They could elect for it to be played as a tape. It is not a tape—it is digital now—but they could elect for it to be played. It is a decision for the court how best to present that evidence.
Quite normally, it is a question and answer process between the prosecution barrister and the police officer. There are risks in that. The way that it is presented could cause an inference. I would expect that a judge or a defence would intervene if it was presented in a biased way or a way that was inaccurate. There would be checks and balances within the courtroom itself.
To answer your question, there is a traditional route. There will be reasons of efficiency and the specific evidence to cover in a court, but there are some checks and balance within that. Yes, there are also potential risks, I would say, in how it is presented. It could be misinterpreted, as other people are, in effect, providing the dialogue rather than what was originally recorded. Does that answer or help?
Lord Blencathra: It throws up lots of other questions, which I will not ask right now.
Richard Kempshall: I would break the reasons that they are read out and presented under three distinct headings. The first is court time. We understand that it is valuable, and we need to look at the efficient use of court time.
The second is the complexity and structure of police interviews. Particularly in complex crime, these interviews can run to days and certainly hours. Even when officers adhere to fairly tight topic control in their interviews, the real key evidence can sometimes be spread out across the duration.
That brings us on to the third heading—the technology available to distil that product and present it before the court has historically been limited. Editing audio cassettes and VHS tapes was incredibly difficult. Now we are moving towards digital technology, the ability to distil that extended complex interview structure down to usable evidence is improving. You have the additional bolt-on of the legal interpretations of when interview material becomes relevant in a trial.
That is how I would explain how and why it is used. The superintendent has touched on the risks. Again, I would break those into three headings. The first risk is the information put before the jury. If we want to make good decisions, we need access to good information. By distilling an interview down to only black and white words, we are denying the jury access to several other strands of good information that would assist its good decision-making. Equally, jurors are specifically directed by the judge at the start of the trial to bring their life and world experiences with them to their decision-making. As human beings, we develop the ability to understand tone, volume and intonation long before we understand words; it is one of our first life experiences. We are denying the jury access to that.
The second risk is around the rights of the defendant. It is often cited that we have an enshrined right to silence in the UK. You also have an enshrined right to speak and to have your words played in your defence when you are at trial. By doing this, we are often denying defendants the ability to have his or her own words in their own voice put before the jury.
There is also the well-trodden path of biases in transcripts and the fact that errors are often compounded on review rather than necessarily cleared up. I experienced that just last week in a trial in the UK with multiple errors in transcripts that took some time to review.
Ben Stephenson: There is not much more to add. Colleagues have very adeptly addressed the question. In addition to that, to build on what DI Kempshall has said, I find it somewhat perverse and perhaps bizarre that the reading out of the defence interview is done by the prosecution with the OIC—the officer in the case. There is a risk of underemphasis or overemphasis in the language that is used, particularly when the officer in the case might not have been the interviewing officer. That is quite common.
You have a couple of degrees of separation between the interview and it then being read out in court. Those two people who are reading it out may not have been present for the interview. There is that lack of quality there. That is the additional point that I would make on that.
Lord Blencathra: I would like to follow up on what Detective Inspector Kempshall said. I will not ask who should be responsible for producing or editing the recordings played to the court. We all accept that no one is going to want to listen to four or five days of interviews when there may only be three or four crucial bits in there. I do not understand why the police have to transcribe the whole shooting match. If you have good-quality digital or video recordings, why can you not send that to the CPS and annotate it to say, “Look at point 52, 78 and 95, and you will find the key evidence”. Is that fanciful or feasible?
To elaborate on that point, if you are transcribing a video recording, how on earth can you build into the transcript nuance or sarcasm? The suspect may give you two fingers. That cannot be played in court. How can you understand nuance or tone? I do not understand how you can do that. Have you had experience of it working properly or not? I am sorry. There were a few questions in there.
Richard Kempshall: In terms of who is or should be responsible for creating the compilation clips, as I call them, I do not know who should be, but I certainly know from personal experience within my unit and my command that it generally falls to interviewers because they have the most direct memory and experience of the interview process. They are often the only ones who know the full content that has taken place in that interview. Where I have presented that evidence at court, it has been either the interview adviser or the interviewer who has created it.
I have personal experience of that working well. A full detailed transcript alongside a specifically targeted compilation of clips can distil a four‑hour interview down to 10 minutes’ worth of key evidence. That can be created alongside the defence case statement, so it can serve the benefit of both the prosecution and the defence.
Can a transcript truly demonstrate the breadth of emotion, tone and intonation in a human voice? My personal opinion is that it cannot, particularly in the realms of “no comment” interviews. They often become agreed facts. They are not even read out to the standard of a transcript. An agreed fact will be created that the defendant was interviewed and provided no comment. It is fair to say that no two “no comment” interviews are ever the same. They should not be overlooked. They should be given the same treatment alongside a talking interview.
While no evidence is provided in “no comment” interviews, Section 34 of the Criminal Justice and Public Order Act talks about inferences from silence. It also talks about inferences from response and reaction. That can be equally powerful in a “no comment” interview. I will be honest. I am at a loss as to how those could be put forward in a black and white written transcription.
Mark Kenny: In terms of the transcript itself, I know how they are produced. In Greater Manchester, we have a dedicated team of professional typists who produce the transcripts. As for whether other people could produce them, that is an option, but they know exactly what they are doing by experience.
One thing to note is that they produce the whole transcript. There are some elements that we summarise in Greater Manchester, but by and large it is verbatim for the main part. If that was edited at the start, there would probably be some adverse comment by other parties: “Why are the police editing it down?”
I would agree with the detective inspector. We produce the full transcript and then we pass that to the prosecution and the defence. The skilled barristers then decide what parts of that evidence will be relevant to the case. In terms of the audio or the written transcript, it can never be as completely accurate—it is quite one-dimensional—when compared to the whole visual and audio presentation of that evidence.
It is difficult. If you go back to the start, the whole purpose of the production of the evidence is accuracy. We want it to be as accurate as possible. From the legal point of view, the court would consider the elements of accuracy and fairness to all parties.
It would very much depend on the circumstances of each interview. There may be things that are relevant that could be played in court as well as having the written document. It could be played. It is all digitally recorded now.
You would also find an argument in court potentially around fairness. The way that people present in interviews can be seen very adversely. That would itself create some conversation with the defence side. If I see it from their point of view, these people are in difficult circumstances. If you can see them, they may present far worse than through the written word.
Ultimately, it is about the best, most accurate production of the evidence. It is for the court to decide what the best way to present that evidence is. Certainly, we provide the interview digitally. We provide it in written transcript, but it is also available digitally to the defence, the prosecution and the court. They would have the option of playing it, should they wish to do so, if there is a controversial point. That is a potential as well.
Playing the evidence is not always as clear-cut as reading it. You probably know that listening to it might be extremely difficult. It is sometimes very difficult to tell what somebody is saying. A skilled transcriber will stop, pause and could take very long time to understand what somebody is actually saying. It is not always a clear transcript or a clear process to produce that transcript. Playing it does not always take you that much further forward.
It is on an interview-by-interview basis. I am sure my colleagues will have experienced this. Interviews can be very difficult to play back and very difficult to play in a court. That may in itself lead into the point about court time and the efficiency of the court as well. There will be considerations around that.
Certainly, that is how we produce it. Those are the key principles for me: the accuracy of the evidence and absolute fairness to all parties.
Q29 Lord Carter of Coles: You raised the point about quality, Mark. My question is about training and support. What do they get? How do you get consistency in it? How difficult is it to do that?
Mark Kenny: That is a really good question. I was reflecting on this on the way here. You touch on a broader point there. Within Greater Manchester, the last time we recruited typists was over a decade ago and probably a lot longer. These are very experienced and very skilled people, but it is not necessarily a skill or skill set that is particularly relevant at the moment in terms of where technology is moving. The people received training, but it was a long time ago. At some point in the past, they received training. We have our own internal manual of guidance, although forces will approach that point independently.
In terms of accuracy and quality, the transcripts are checked by the officer in the case. Once the typist has produced the document, the officer in the case is obliged to check it, exhibit it and provide a statement to the court to say that it has been checked. The other checking mechanism is that we provide it to the defence. The defence solicitor may have been present in the interview, but, if they were not or if a different counsel is present at court, they have access to that recording. If there are differences between the typed document and the evidential presentation of that, there are checks and balances within that, which the defence would be able to pick up on and object to. If there are errors, those will be flagged up.
We get a lot of feedback from courts and the CPS around the quality of evidential files. It is one area that we do not get that much feedback on. We do get feedback on other aspects of file production.
To answer very specifically, they are very experienced people. They are very experienced at typing. They have some awareness of law but not a great deal, so points to prove. Their key focus is the production of that transcript and the particular questions that lead directly to the offence concerned. That is their job, essentially.
Lord Carter of Coles: You touched on having a manual, which is quite interesting, so as those people are inducted, there is a way of doing it. Richard, how does the Met do it? Do you have a manual? Are there standards? Across the different forces, we are hearing that people have a different approach to training and things.
Richard Kempshall: If I may, it may be easier for my colleague to answer that. It is slightly outside my remit.
Ben Stephenson: Thank you, Richard. I have been very lucky to forge a close relationship with our typing services that work in London across the Metropolitan Police. They have Pitman training. It is a two-day training course. They have to reach a target in terms of words per minute for their transcribing skills, but ultimately it is accuracy over speed, because interviews are becoming more complex by their very nature.
The initial disclosure that the police are now expected to serve in an interview is far greater than ever before. There are reams of digital data that we get from phones and laptops; there are streamlined forensic reports, fingerprints and DNA; there is dashcam footage from cars, Ring doorbell footage and body-worn video, which as you know is in the public eye at the moment. We have it on all the time now for evidence gathering. There is all that information that we have to share beyond just asking questions and receiving the answers, whether it is a “no comment” or a talking interview.
It takes typists around a day to type up an interview. That is for 30 minutes of audio. It is a long, arduous process. It can take up to a day for them to transcribe that.
In the Met, we do ours verbatim. We want to make sure the caution has been fully understood and explained. There are reasons for that. The first is if any adverse inference is going to be drawn. DI Kempshall mentioned Section 34 of PACE. You also have Sections 36 and 37—marks, scars and tattoos, being in the place at the time of arrest, et cetera. There is often a conversation that goes on before the questions are asked that is relevant to the investigation. We really make a lot of effort to capture that.
The second point I would make—this is entirely what Superintendent Kenny said—is that it is not an attractive vocation at the moment. The vast majority of our typists in the Met have been in our organisation for 15-plus years. Recruitment is difficult. There is a long vetting process behind that. Because of that, they may find another job that they would prefer to go to. To be honest, from an objective perspective, if it takes up to a year, why would someone want to wait a year to start a role? As you know, vetting is in the public eye in terms of how long those processes take. It has to be more in depth because of the sensitive intelligence information and evidence that we have in our possession.
In terms of ongoing training, I spoke to senior colleagues in operational support services. They have regular feedback sessions to make sure the typists understand slang words and in words. Again, London is a melting pot of languages, linguistics and semantics. For a lot of people being interviewed, English is not their first language. It might not even be their second language. We have that to contend with.
The final point I would make is that they are exposed to a lot of trauma; they are not police officers. We are trained to deal with that trauma; they are not. Again, there may be things that they have heard or seen in the interview that make them uncomfortable. There is a pastoral element that is delivered to them to make sure they are able to cope with the information that they hear. If every single interview is transcribed by them—the vast majority are; we have to outsource a few when our resources are overwhelmed—they are going to be listening to some really quite horrible comments and information, and seeing things that maybe they have not been trained to deal with beforehand. There are sessions afterwards to make sure their concerns are listened to and addressed appropriately.
Lord Carter of Coles: Finally, are we getting a sense that there could be a workforce shortage here? Are you feeling that?
Ben Stephenson: Absolutely, yes.
Mark Kenny: It is a risk going forwards. As I said, in Greater Manchester, we have not recruited recently. Fortunately, we have not had to. The typists that I am responsible for have 30 years’ service. That is quite a long time. They are very dedicated.
There will come a time when we get more retirements. It may become an issue in the future simply because typing, as a skill, as a vocation, might not be the most desirable at the moment. Times have changed. There probably was a time when typing or word processing, some time ago, was something that people did training on at college. There probably still is training for typing, but as time has gone on, it has become less relevant.
The Chair: Interestingly, you are obviously retaining your staff.
Lord Bradley: I have a very quick question. You were talking about nuance between the audio and transcription. Where there is an appropriate adult supporting a vulnerable person, does that get transcribed in some way so that the nature of the responses reflects their vulnerability and the nature of the communication as a result of that?
Ben Stephenson: Yes, absolutely. For all parties present, whatever is said is transcribed. It is particularly important that you raise the appropriate adult. Again, when a youth is interviewed, there must be an appropriate adult present, unless it is an urgent interview. That is governed by PACE, with the Criminal Justice and Public Order Act being sort of secondary legislation to that.
You are exactly right. It is about making sure the appropriate adult is there in order to explain or clarify a question that has been asked, as an intermediary would at trial for someone who is vulnerable, intimidated or under special measures. They are there to make sure the questions are relayed appropriately. Again, that will be part of the transcription. It has to be there. I do not think a judge would be comfortable directing a jury to draw an adverse inference if they were not certain that the suspect who then becomes the defendant understands exactly what is being asked of them.
Mark Kenny: It can also be quite complicated. You can have a number of parties present. As you probably know, during an interview clearly you have the defendant, but you also have potentially the solicitor, an appropriate adult and possibly an interpreter, and then you have the police officers. There can be a lot of parties present and a lot of people talking. Part of the skill of the transcript is to determine who is speaking and the detail because, again, that is not always clear.
If you were to play a recording, that is something that you would want to know. Who is speaking at what point? You would probably need a transcript and a recording, thinking about it, in cases such as that. Clearly, it would be far more complex, listening to that recording, to know who is saying what at what point. It can become more complex in that scenario.
Lord Carter of Coles: You are two major forces. I am just wondering about the quality of it in smaller forces. Do you have any experience of that from having worked anywhere else?
Ben Stephenson: I have only ever served in the Metropolitan Police so I cannot answer that question, I am afraid.
Mark Kenny: I have only worked in Greater Manchester.
Q30 Lord Blencathra: As far as Greater Manchester and the Met are concerned, some witnesses have suggested that, while the courts have lousy playback facilities, some police forces have lousy recording facilities. They have a little camera at one end and a microphone at the other. People are sitting there mumbling and the recordings are not good enough. I do not want you to drop Greater Manchester or the Met in it, but are you generally satisfied that your two major polices forces—you are the largest ones in the country—have, in most interview suites, adequate sound and video recording stuff?
Ben Stephenson: I am satisfied. The other thing that matters is when you are doing interviews in prison as well as interviews at the wicket when the defendant refuses to leave their cell. We do have portable video equipment that officers should be using on every occasion. That is understood to have exactly the same software as a fixed terminal in an interview room in a custody suite, irrespective of whichever custody suite that is across London.
Mark Kenny: I am happy with what we have. It is new. It has been replaced recently. The recording equipment is in all our custody facilities. It is digital. Whether it is clear is partly down to people speaking, of course. There is always that. The playback is not always clear, but the technology is new. It is there. We digitally share the interviews as well. We have a mobile equivalent, although most interviews will be in police stations. We have the technology.
The Chair: Is there no call for a national framework? Have you and your colleagues in different forces ever discussed whether it would be a good idea to have a national framework in this area of policing?
Ben Stephenson: I do not believe there has been a national agreement. There has certainly been no national agreement on the standardisation of transcription.
The Chair: No, there is not. You do not think that would be helpful in any way.
Ben Stephenson: It would definitely be helpful, yes. The NPCC would probably have to be involved in that conversation with the College of Policing to decide on a policy framework. I do not know whether legislation would be needed. That would be something for someone a little more senior than I am in my current rank.
The Chair: You are doing the job. You think it would make the job more efficient, easier to do or more effective.
Ben Stephenson: I think so. It would require all partners in the criminal justice system to come together and agree a policy framework for that to take place.
Mark Kenny: I would agree. There is an opportunity for some standardisation. Greater standardisation nationally in this area would be helpful. I would not say otherwise. We can only talk for our forces. There will be more commonality between forces, I suspect, but there will be some variations around the country. It would be sensible at least to look at standardisation as an area to explore.
Q31 Lord Shipley: Can I thank everyone? It has been an extremely helpful set of submissions so far. I have two general questions and then two specific follow‑up questions to individuals arising from what has been said. First, I would like to dig a bit deeper into the quality of transcripts and who is responsible for ensuring the highest quality. Secondly, considering the level of quality assurance in place in respect of your own forces, how effectively do your transcripts reflect the interviews or forensic audio?
The first specific follow‑up question is to Detective Sergeant Stephenson, who made the important point earlier that the people who read out transcripts may be two steps removed from the originating of the transcript. I wondered whether any research has been done on that. Are there any examples of potential errors being made in the system as a consequence of the person reading them being two steps removed?
Secondly, Superintendent Mark Kenny made a number of statements about the accuracy and quality of transcripts and said they could never be fully accurate. I understand the point that is being made. We have just been looking at whether there should be a national framework. Superintendent Kenny, do you have any knowledge of any examples of police forces working together to assess the quality assurance of what they are doing? That can be done informally. It does not have to be part of a framework; it can just be good practice.
Let us start with Detective Inspector Richard Kempshall. I would like you to address how we assure quality.
Richard Kempshall: My experience is that, when the transcript is generated, certainly with my teams, it is incumbent on the interviewing officer to review it and make sure that they are happy with the content of it. There are occasions within more volume crime work where it is the officer in the case, not necessarily the interviewing officer. In specialist and complex crime, it will generally be the interviewing officer.
You then have the additional layer of your disclosure officer, who will, in most cases, review all the material that is submitted to ensure that it is relevant and accurate, and that it is not undermining to the prosecution case or assisting the defence. You then have the CPS layer and then finally the trial barrister.
All that being said, I have regularly experienced being at court going through transcripts and still finding errors. Where there is a problem with review is that often errors are compounded. When one person hears one thing, it then becomes compounded and everybody who reads it afterwards hears that same problem. Another issue is that, often, different people will hear different elements, and they will all interpret it slightly differently.
This is why I am probably a strong advocate that this should not be an either/or. There is a place for the transcripts, with an acceptance that they may contain errors and others may hear things differently, to be read alongside the playing of the original product as a supporting document rather than as a replacement document. It should not be an either/or. That would provide the most effective safeguard to any errors that might be in the transcriptions.
Ben Stephenson: I would largely agree with DI Kempshall. The only difference that he and I have is purely our history in the Met. He has been in specialist crime; I have been in volume crime, whether it is CID, burglary and robbery squad, or where I am now. The interviewing officer often will not be the officer in the case purely because of shifts, et cetera. The time it takes between the interview, the case file being prepared for charge and the CPS authorising the charge and requesting the ROTI could be a number of months, if not years, depending on the genesis of that investigation and how complex it is.
The Chair: The interviewing officer is the person who did the interview. Who is the officer in the case?
Ben Stephenson: The officer in the case has conduct of the investigation. They are the person who will be at court. They will be the support to the prosecution barrister. The interviewing officer, apart from doing the interview, may have no more involvement in the case. That is particularly pertinent in the Metropolitan Police because of the sheer size of our organisation. Investigations may be passed to another department or maybe even another BCU, depending on the primacy of the offence and where that investigation is going to sit.
This is what your Lordship rightly raises: you may be one or two steps away from the interview. The officer in the case was not in the interview room and does not have the memory of being there. They might not be able to check it and go, “That is not exactly what they said” and refer to the transcript. The officer in the case is looking at that transcript as blind as anyone else in this room who has not been present in that interview. There is a risk around that.
It is the responsibility of the officer in the case to rigorously review that transcript, and the risk is even greater when they were not present at the interview. Saying that, I am conscious of the time that officers have in the organisation at the moment. The number of case files and crime reports that officers are carrying now is pretty high. It is very difficult to resource everything that is expected of them in order to make everything satisfactory for the trial to hear when the evidence is entered.
At the end of the day, they are signing and exhibiting a statement to confirm that what is being provided is accurate. There is a personal responsibility and obligation to make sure that what you are submitting to the CPS is correct. That is what I would say on that.
Mark Kenny: I agree with all the points that have been made. It is the same in Greater Manchester in terms of that process of validating the document, as it were, through the different parties.
There is a risk of human error. There is no doubt about it. We are reliant on the officer checking and ensuring that the document is accurate. They may make an error, but, reassuringly, there are other parties. The defence should be the first people to challenge if there is a police error. There is then the CPS as well.
There are checks in the process. Are there flaws? Yes, but equally, the fact that there are checks and balances ensures that the product of the transcript is typically the best it can be. There can be differences; there can be misunderstandings.
How the evidence is presented in the court is difficult. I agree with Richard to a point, although my take on it would be that you would choose whether to rely on the transcript or the recording on a case-by-case basis. The court may be better informed by both, but, equally, it may be the case that the court decides that one is sufficient. It would have to decide that on an individual basis. Equally, the recording may not take you any further. It is difficult to say on a general basis.
The difference from now might be that the court would have more of an opportunity to decide that it should be the recording and the transcript together. Maybe that is the difference from now.
Certainly, I agree that reading it out could be flawed. As I said before, if it was done particularly badly, I would expect that to be challenged by either the judge or the defence barrister. The barristers in court are particularly skilled at getting to the nub of the evidence. Ultimately, that is their role.
To answer the other question that your Lordship asked before around forces working together in this area, I will be honest. I am not aware of forces collaborating in this area. Forces work regionally with the CPS, for example. There is quite a good footprint in terms of that. As you probably know, the CPS is regionalised. Forces have joint operational meetings with the CPS on a regular basis, where we look at the production of evidence and make sure that we are following best practice and working together on the best and the fairest presentation of evidence. We work very closely with the CPS.
In my experience, transcripts is an area that has not come up. Certainly, in my span of experience, this specific area has not been mentioned. Other areas certainly have. The quality of other parts of files or the production of evidence does come up quite a lot, but it is not so much transcripts. I will be honest. I am not aware of forces working regionally on this as a specific thing.
Ben Stephenson: Do you mind if I make one more point? To go back to this point about the checks and balances that are woven into the process, there is an even earlier possible check. Let us say you are the interviewing officer. You ask the suspect a question and they give you an answer, but the answer is ambiguous or vague, or potentially has the opposite meaning of what they are trying to say. A particular example is, “I didn’t do nothing”, which is common lexicon in certain parts of London. I have been in many interviews where that has been said.
I would encourage the officer to have the confidence to clarify what that person has said, so they are giving the evidence that they want to give and not saying something unintentional or completely perverse to what they are trying to raise. That is really important. We could do that with any interviewer, whether it is a police constable, detective constable or beyond, to make sure they are confident in challenging the suspect and going, “I didn’t understand what you said. Do you mind explaining it again?” or, “I think I know what you said, but I’ve got to try to understand exactly what you said so that I can represent what you have said fairly and objectively potentially to a jury”.
Q32 Baroness Pidgeon: That is very interesting about the use of language and how language changes. You are absolutely right. My 12 year-old regularly says things that I have to get explained to me. I want to pick up the use of AI technology and voice recognition. We had some evidence from the CPS to say that it has considered the use of AI or technological innovation for smart transcription and it has real potential. There will have to be some safeguards around data protection, accuracy and so on, but the CPS thinks it has potential.
We had some other evidence from one of our speakers last week, James Tompkinson, who said that five police forces that they are aware of have started using automated transcription software. Perhaps I could start with you, DS Stephenson. What potential is there for technology and AI to help in this area and help in the future with the potential shortage of specialist typists? What are you seeing the Met using in this field at the moment?
Ben Stephenson: I am going to defer to DI Kempshall because this is very much his area of expertise. If you do not mind, I will let him answer.
Baroness Pidgeon: I will go over then. From what I heard at the beginning, I thought it was you. DI Kempshall, can you help on that one, please?
Richard Kempshall: Yes, of course, largely because we are running some simple trials informally within our units on specialist crime around the use of AI transcription and the potential value that it is going to add.
What we are seeing at the moment, in very simple terms, is that the period of time that my colleagues mentioned, a day or so for 30 minutes’ transcription, is greatly reduced with the use of AI. We are also seeing, however, that the number of errors in AI, in that shorter period, is higher. The product is reaching our officers to review more quickly, but it contains more errors than the human-created transcript.
There is potential, but it is by no means there at the moment. It can bake in bias. It can interpret. It struggles with multiple speakers. We have seen it add a third person to a conversation where they have not been there.
There is definitely potential. With that being said, whether a transcript is human-created or AI-created, it is only as good as the clarity and the quality of the interview that has been conducted and the accuracy of the subsequent human review. You still need the human at the beginning to conduct the interview and the human at the end to review the product, whether it is computer-generated or typist-generated.
Before AI, the biggest relevant evolution for interview technology was the ability to digitally record and the ability that now gives us with editing, cutting, clipping and creating compilations. I feel AI has potential, but the recent move to digital recording has been the biggest technological advance for us in interviewing and the presentation of evidence.
Baroness Pidgeon: How recent is that move? We questioned earlier whether you were still, in effect, cutting tapes, and we were told that everything was digital. How recent has that move been?
Richard Kempshall: DS Stephenson may correct me, but large-scale rollout within the Metropolitan Police was within the last two years or 18 months.
Ben Stephenson: Yes, the DIR has been in place since then. The custody suites were done suite by suite.
To supplement what DI Kempshall has said about using AI to transcribe DIR for the purpose of ROTI, we are not allowed to do that outside of specialist crime within the Metropolitan Police. It is very much used as an informative piece for the interviewer to build a summary of the interview, which would then be presented to the CPS pre charge as a short descriptive note to explain what has gone on in the interview. That cannot be relied upon for the purposes of transcription. We still have a human hand, which is the typist.
Baroness Pidgeon: It is a summary. So you are using AI to help with the summary in this trial. Has that proved to be useful?
Richard Kempshall: DS Stephenson is correct. Within the wider MPS, we are not using it full scale for transcriptions. On specialist crime, we are running a few very informal in-house trials, in which we are trying to generate the full transcript rather than the summary. These are very tightly controlled. They are informal and in-house. They are largely at the moment to advise us on the capability.
Within specialist crime, we have the luxury of slightly more resource than the wider MPS, which gives us the ability to focus some officers on spending that extra time to go over the errors that come out of the AI. It is largely informal and informative at the moment.
Baroness Pidgeon: You have not used that. Were you saying that it has been used, or have I completely misunderstood?
Ben Stephenson: For the purpose of transcription, we cannot build the MG15, which is the ROTI, from the DIR transcript, but it can be used by the interviewing officer to review what has been said. They can then clarify that and provide it to the CPS for charging advice.
Baroness Pidgeon: Have you been using that?
Ben Stephenson: There is nothing stopping officers from doing that in order to build the case file for charging advice, but once it has been charged you would certainly not be able to rely on that for trial. There is a differentiation between the two parts.
Baroness Pidgeon: What about Manchester? What are you doing?
Mark Kenny: We are not using it for the production of transcripts. The technology is moving really quickly. I am sure it will improve, but when we have looked at it—we have looked at it recently—we felt there was a high error rate with the use of AI. For example, regional accents are a factor in that. There are some diverse regional accents within Greater Manchester. It is a serious point because it is around the accuracy of the product, the MG15, the transcript. We are not using it for that very specific purpose. AI is used in some areas. It is starting to be looked at but not for the production of transcripts at the moment. In the longer term, yes, potentially, but it will still be important to check it.
However it is produced, it must be checked and validated. That has to be absolutely key. Certainly, as it stands, we did not think it was accurate enough to produce the document.
Richard Kempshall: This slightly dovetails into a point that I have not raised yet, which is how we view the purpose of a transcript. If the purpose of a transcript is to be a reference document that is easy to read through, easy to annotate, easy to mark, and can be easily passed through the different levels of the prosecution system, transcriptions, whether they be human or AI, are probably quite suitable at the moment.
If we view a transcript as original evidence of the interview to put before a court, we will always come back to the fact that it will have baked-in errors regardless of whether it is human-created or AI-created, in which case the interview footage will always be the best-quality evidence.
We may somehow over time have blurred the lines on the purpose of a transcript. I am not quite sure where that line should be drawn. Is it a reference document for research and informative value or is it pure evidence to be put before a court when considering a prosecution?
The Chair: It fulfils more than one function, does it not?
Q33 Baroness Cass: We heard previously about quietly spoken interviewees, people mumbling or whatever. Do you ever mic people with a lapel mic or any other body mic, which would also help in distinguishing the interviewee from other people in the room? Following on from that, would that be easier for AI to transcribe? Have you considered or do you use AI as a facility for the transcribers to speed up their own process rather than going directly from the AI transcript to the officer?
Ben Stephenson: On the first point, I am not comfortable with attaching a detachable piece of equipment to a suspect in an interview room because I do not know what they are going to do with it. If they are looking to get out of the interview room, they could put it in any particular orifice on their body and then we would have to rush them off to hospital. That obviously takes up a lot of resources. That is not feasible.
On AI, I cannot answer that point directly, but, because of my very good relationship with our operational support services around typing, I can certainly go back and ask them whether there has any been any conversation among peers, colleagues and their senior leadership team about whether they are starting to rely on the AI transcription of the interview, where the DIR has been updated, to see whether that can assist them. I would have to take that away and find out, I am afraid.
Baroness Cass: The interviewee may be a complainant, not a suspect.
Ben Stephenson: I thought we were talking specifically about suspect transcriptions. Certainly, video-recorded interviews with complainants take place in dedicated suites. When we are looking to apply for special measures, whether it is Section 27 or Section 28 of the relevant legislation, we have a dedicated suite for that to take place. There are microphones fixed around the room with visual equipment. That evidence can be entered as evidence-in-chief, as a VRI, if a judge agrees that special measures apply. That position is satisfactorily covered at this stage.
Q34 Lord Blencathra: I have a few little questions, Chair, if I have time. If your SOCO forgets to put on his plastic booties and he walks over the murder scene, that evidence may be discarded because there is a fault there. If there is a slight glitch in the DNA—it is handed over and not signed for—it will be discarded. It worries me that there can be misinterpretations of words in the transcript and that it does not take account of the nuances, but that evidence does not seem to be discarded.
DI Kempshall, I am old enough that in the 1970s, in order to get my favourite music tapes, I had to sit there with a razor blade and a little splicing gadget to join up the tapes. That has largely changed now. One witness said last week that the CPS or the lawyers were trapped in a 30-year time warp, where they thought reviewing the evidence would take hours and hours of playing back audio tapes.
Would you agree that, where forces have the technology and the audio is good enough, the best available evidence must be, possibly as you have suggested, to have a transcript of the relevant bits and that the relevant bits are also played in court? That is a loaded question, but I am going to ask it. Do you agree?
Richard Kempshall: Speaking as an interviewer, yes, it is a loaded question, but I do agree. Speaking as an interviewer and as somebody who has acted as a senior investigating officer at trial, I know that how a barrister uses the evidence will always be their decision. It is their forum; we are in their courtroom. It is incumbent on us as investigators to provide them the best evidence in the best format to allow them to use it. I personally think that, with the new technology that we have, and the ability to create very simple compilations and to provide those alongside the written transcript, that is the best possible evidence that we can put forward.
As an aside to that, something that we have not commented on is the quality of the interviews in order to make them admissible in the first place. That requires a level of training. We invest in that training nationally across the country to ensure that our interviews are of good quality and they do not fall to be inadmissible because they are poor interviews.
If we can put together the training, the quality of the interviews and the structured compilations alongside detailed transcripts, we are providing barristers with the best opportunity to put the best evidence before a court.
Q35 Lord Blencathra: I was Police Minister with Michael Howard in the Home Office from 1993 to 1997. One of our concerns was that you can have excellent bobbies and excellent investigators, but interviewing technique was a rather specialist skill. Some of the police were not up to speed on that.
I am coming at this from the fact that I introduced the legislation on twin-deck recording. Michael Howard and I both naively believed that that was it and the recordings would be played in court in every circumstance. I am appalled that we are in the transcript business.
I have one final question. Some lawyers have said that they are not aware of any potential miscarriages of justice. I simply do not believe that prosecution and defence barristers have played back the whole recording and read the whole transcript to determine whether it is accurate. They say that they do, but I am sceptical enough not to believe that. If you have not looked for them, you might not find miscarriages of justice.
You have hinted that possibly there were some cases where the evidence read out in court was not consistent with the tapes. Are you concerned that there might have been potential miscarriages of justice?
I have one final question to all of you. If you were starting from scratch again, what system would you invent for getting this evidence into court?
Mark Kenny: In terms of miscarriages of justice, it is impossible to say. We present the evidence as best we can. We hand that over to the parties. It is for them to use it as they see fit within the courtroom.
We provide the digital recording now as well as the transcript. They use the transcript but perhaps not the recording. DI Kempshall is right that providing both is probably a better process, but I hold with the point that it should be case by case. It depends on the circumstances and whether it is felt necessary, in terms of the best presentation, the accuracy and the fairness of that evidence, to present both. It is case by case, but we are presenting it as best we can.
Have there been miscarriages of justice? I cannot comment on whether the defence has checked it. It should do. That is what its role is. The CPS should check it. That is what its role is. All the parties involved should check it. I cannot comment on whether they do in every case. I am sure some people will not. We provide it. It is there. The defendant would probably also object and say, “I didn’t say that” to their counsel.
I am satisfied that we have a system. Yes, there are potential flaws, but we have checks and balances. It is not just the police or the prosecution; it is the defence. All parties are present, and they have the opportunity to object and view that evidence. We are in that position.
Of course, tape recording and now digitalisation is one step from that. That brings us into the modern era, really, with provision by email. We provide it to the defence. We send them a secure link to the interview, so they can listen to that in their own time and so on. We provide it efficiently in that regard, but I cannot say there has never been a miscarriage. I would hope not, because we do our best to provide accurate evidence.
How should it be done? I hold with the point that there is no perfect answer to that. We record it. Of course, the scrutiny, correctly, is on the police to provide the evidence that is admissible. You asked, “If you were to start from scratch, how would you provide it?” Certainly, the position that we have now is a good one in terms of digitalisation and the presentation of evidence as best we can, with all the caveats that there are. There are potential pitfalls, but we are providing it in a very efficient way.
There is another option—again, this would be very case specific and subject to some discussion—which would be the video. You would have audio and video as well as written. As we just mentioned, for vulnerable victims, routinely and correctly, we take video evidence in evidence suites. That is considered the best evidence for children in particular and for adults in terms of communication. In fact, that is the absolute ideal presentation of that evidence.
Should that be a way of providing defendants’ evidence? It remains an option to provide that evidence as well, if forces have that capability. Again, it would have to be on a case-by-case basis. It is not a one-size-fits-all. That remains another option that could be considered in the future, but, again, in the interest of fairness, there are probably a lot of people who would say that providing the video would potentially prejudice that evidence as well because the way that people present can be seen very negatively or, some could say, unfairly.
There is not an easy answer, but all these things would have to be considered in the presentation of that evidence for it to be absolutely accurate and fair.
Q36 The Chair: I am going to incorporate this with the last question. If there are any particular recommendations that you want us to include in our report, could you include those in your answer as well? If it is all right, I will take your last comments as some of the things you thought might be recommendations.
Ben Stephenson: My view is that the system, as it is, is largely satisfactory. We need to embrace AI and the benefits of that, with the strong caveat that we have that human hand and a strong regulatory framework to ensure that the evidence is produced and provided to the best quality.
What Superintendent Kenny said about visually recorded interviews being already offered to complainants or witnesses who are vulnerable or intimidated under the relevant legislation is entirely right. There is an argument to offer that to suspects. There is a policy in the Metropolitan Police Service where we routinely do that, particularly when a male officer might interview a female suspect, because there may be allegations that are made. If you record the interview visually, you can rebut any allegation that is made about the conduct of the officer in that interview. It is all about safeguards against criticisms or complaints that may be made against us. Body-worn video addresses that a lot as well. As well as an evidence-gathering tool, it is a way to rebut public complaints.
I am trying to be as succinct as possible and conscious of time, but what Superintendent Kenny has said is entirely right. I would just make sure there is a regulatory framework in place to ensure robust oversight of transcriptions.
Richard Kempshall: I am sure my view is not going to come as a surprise. The purest form of evidence is the best evidence. It is not a case of one or the other. They should be mutually supportive, because the purest evidence and the transcript have different purposes in the criminal justice system. The one overriding safety net to all of it is the fact that the defendant tends to go second at trial. They always have the ability to rebut any errors that come out in the prosecution evidence.
I believe we should be using the video and/or audio in support as often as possible, as the best possible evidence to put before a fair trial.
The Chair: That is helpful. Before we close the meeting, this might be a nod of the head, but have you been talking about both magistrates’ court and Crown Court? Would it have been a different conversation had we been talking about magistrates’ courts?
Ben Stephenson: It would be a slightly different conversation about magistrates’ courts, because police transcripts are not routinely provided for a magistrates’ court trial because there is a bench. It would be for a judge and jury at Crown Court.
The Chair: You have been, in essence, talking about Crown Court.
Ben Stephenson: Yes, pretty much, the reason being that we do not produce ROTIs for magistrates’ courts unless they are required in specific circumstances. Having said that, for Crown Court and youth court, you would, because there could be an either-way offence or an indictable offence heard in the youth court that would take place in a magistrates’ court. That would be the only exception where a ROTI would be provided. It would be for Crown Court and youth court.
The Chair: Okay, we will look at that further. That is very helpful. We will close the meeting there. That has been exceptionally helpful and interesting. We appreciate how busy you are. We are very grateful for the time that you have given us today.
