15
Examination of witnesses
Fiona Rutherford, Sam Skinner and Harpreet Sandhu.
Q13 The Chair: Welcome to this meeting of the Public Services Committee and our inquiry into police transcription. We very much welcome our witnesses from the legal profession today to give us evidence. I would like to start by asking you to introduce yourselves.
Harpreet Sandhu: I am a criminal practitioner. I prosecute and defend. I am the current leader of the Midland Circuit and I represent the Bar Council for today’s purposes.
Fiona Rutherford: I am the chief executive of Justice, the law reform and human rights charity.
Sam Skinner: I am a criminal barrister. I also practise in public law. I work in the Midlands, where I prosecute and defend. I am also a Crown Court recorder, but I appear today to give evidence on behalf of the Criminal Bar Association.
The Chair: You are very welcome.
Q14 Lord Bradley: Good morning to you. This is a general opening question, but I may follow up with two more specific questions. How are police transcripts used in court and how important are they to the delivery of justice?
Harpreet Sandhu: We wonder whether we might answer all of that by providing you with a worked example of a very short trial process, which might give a flavour of how transcripts come to be in existence and how we then deploy them. Sam and I wondered, together with Fiona’s help, whether we might talk you through a very short murder trial.
Let us say, for example, there is an allegation of murder and one of the prime witnesses for the prosecution is a vulnerable witness. That vulnerable witness’s evidence will be taken, in the first instance, by way of a recorded interview, which is what we know to be an “achieving best evidence” interview. That interview will be video recorded. That video-recorded interview will be transcribed. Ordinarily, the process of transcription for that interview is outsourced by the CPS to an external organisation.
That transcript, when we get to trial, will be useful for the purposes of case preparation. The reason that the transcript is so helpful for case preparation purposes is that it allows us—that is, those who appear in the case—to identify evidence that is relevant or irrelevant, evidence that is admissible, and evidence that is repetitive and does not need to be shared with the tribunal of fact ad infinitum.
As we proceed towards trial, we take that transcript and identify the material that should and should not remain. We use that as the basis for the recording being edited before it is adduced in court. Once we are at the trial, that transcript, in its edited form, which matches the edited recording, will be used by practitioners to follow what is going on, but also by the judge. Having it in writing allows the judge to summarise that evidence in the summing-up in an effective way.
There is another reason why those transcripts are important. We should bear in mind that the court estate does not have attached to it material to operate at an optimum level. I am afraid to say that that also relates to the playback equipment. It is our experience that there are occasions when the quality of that equipment is insufficient to allow any recording to be heard fully by those who need to listen to it. In those circumstances, we would provide the jury with the transcripts, in edited form, of that witness’s evidence for their evidence in chief. Once the jury has heard that evidence, the transcripts are taken away, so that there remains parity between that witness whose evidence is recorded and given, and evidence of witnesses who come to give live evidence. In those circumstances, the transcripts are used.
Even if, however, the playback equipment is of a sufficient quality for somebody who is speaking loudly to be heard, we must bear in mind that there are some witnesses who are, for example, quietly spoken. Those quietly spoken witnesses will not have their evidence heard in same way as, for example, if I were giving evidence. In those circumstances, we therefore use the transcripts to assist.
A short answer to the question you asked is that the transcripts are used because it assists the process of receiving that evidence. That is one way in which we use transcripts in what we consider to be an effective way, in terms of both preparation and presentation.
I have touched briefly there on the responsibility that is taken for identifying evidence that is to be placed before the jury. We understand that this is a matter with which this inquiry is concerned. The responsibility for which evidence is placed before the jury rests with those who are charged with presenting the case, whether it be for the prosecution or the defence. There is a degree of co-operation between both sides to agree what should be placed before the jury.
If there is any controversy and in the situation that sometimes arises where there cannot be agreement because, in an adversarial process, we may have differing views that cannot be elided, it is for the judge to determine what goes before the jury. A judge will determine, “This part of the recording should be played. This part will not”. It does not always need to happen because, as I say, notwithstanding the fact that it is an adversarial system, the system works well because the professionals engaged in it co-operate where they can and to the degree that they can.
That is one way in which we use transcripts. I will ask Sam to take over and deal with what happens when the evidence of that witness is used to challenge the account of a defendant, for example.
Sam Skinner: One might imagine, with the witness having been video interviewed and given their complaint by recording, that the easiest way to present that evidence at trial would be to play it to the jury, but that is not how things work in practice. As Harpreet has said, the quality of the video recording equipment, for example in the police stations, in serious sex cases at the sexual assault referral centres, and even in courtrooms themselves when there is a Section 28 procedure—that is a video‑recorded cross-examination—is low. Looking round at the recording equipment here, the quality within police stations is much lower than it is in this room. You would be better off getting a 12 year-old with an iPhone to video record people giving their complaints.
Not only is the quality of the equipment low, but the Home Office provided a long and helpful guidance document, the achieving best evidence guidance, with an appendix explaining to the police how they ought to set up cameras and microphones. That appendix and those instructions are routinely ignored. Cameras are set up in the corner of a big room like this looking down to the top of a complainant’s head. Often the microphone is in the other corner. As Harpreet says, when one has a softly spoken witness, they turn up the gain on the microphone, instead of moving the witness or the microphone closer, so you get a really unhelpful buzzing along the whole of the recording.
One gets to a situation when being briefed, for example, to prosecute a case, a murder perhaps, and we have a vital witness or a vulnerable complainant, maybe a child or somebody with mental health difficulties, where they are softly spoken and you cannot really hear what they are saying on the video. The camera is far away from them, so you do not have a good view of their face. You are relying therefore on a transcript of what they said to prepare the case, work out how one is going to present the case to the jury, identify what can and cannot go in front of the jury, and identify how the case will be challenged by the defence. Then the transcript becomes vital.
Even if one has a particularly good recording, the facilities for chopping, changing and editing video do not exist yet in the court estate. One cannot simply cut out anything that is irrelevant and move to the next passage easily. It has to go back to the third-party provider to do that work. We do that by editing the transcript. That is the only way we can get the information back to the third-party provider.
It is not just witnesses, but suspects as well. When suspects are interviewed by the police before charge, often they will be video recorded during those police interviews. Always, they will be audio recorded. It is the transcripts of suspects’ interviews that play an equally large role in the trial process. Suspects have a choice: they can answer “no comment” or give an account of what they say happened. When suspects go “no comment”, you might think to yourselves, “We don’t need a transcript because they’ve gone ‘no comment’”. Actually, we do need an accurate transcript.
We use transcripts when suspects go “no comment” at trial, because the right to silence was modified in the mid-1990s. That means that the jury, in certain circumstances, can hold it against a defendant if they kept quiet in their police interview and went “no comment” but then give an account before a jury. In order for the jury to draw that adverse inference and to hold it against a defendant, the jury has to know what the suspect was asked by the police during the interview. If you do not have a transcript of what they were asked, there is no short route into working out whether it is a fair decision for a jury to hold against a defendant something that they did not mention in the police interview.
Lord Bradley: That is a very clear and comprehensive explanation, so I am grateful for that. I have one point to pick up from it. Having experienced myself the lack of quality in police stations in the recording process, is there a check on the transcript against that poor quality? If you are relying on the transcript and, because of the quality in the police station, it has not accurately pulled out all the information required, is there a check back on that?
Sam Skinner: The difficulty that we encounter most commonly with transcripts from video-recorded interviews is where the transcriber cannot hear what the witness has said, so the transcript will say “inaudible” or something like that. There is a process within the CPS where the reviewing lawyer within the CPS will have had a chance to watch the video with the transcript in front of them.
By time it gets to Harpreet or me, as prosecuting counsel, we will, invariably, sit and watch the video with the transcript in front of us. It is at that stage we begin—I am sure Harpreet would agree—to say to ourselves, “I can work out what that witness is saying”. Then, as Harpreet said, because both sides in our system tend to work well together, I can talk to defence counsel or, if I am defending, to prosecution counsel. Between us we can amend the transcript and make sure that there is a check that way.
Equally, when I am defending, the solicitor who instructs me, who often, for various reasons connected to finance, will not be able to come to court for every day of the trial, will have a chance to check the transcript as well before briefing me. There are checks within the system.
Harpreet Sandhu: We should say that the recordings are routinely made available. There was a suggestion I think I read in the transcripts from your previous evidence session that that does not always happen. That certainly is not our experience.
Q15 Baroness Cass: You have made a very compelling case on the importance of transcripts and why their utility is crucial. Conversely, what risks do you think there are to the current approach to transcription in delivering justice, particularly given that you can lose nuance during transcription and then, when it is reread in court, somebody can put a completely different nuance and emphasis on it? What do you see as the risks? Particularly, do you have any examples where you have concerns that the transcript has not served either side well?
Harpreet Sandhu: We have given quite a lot of thought to this issue, again having read the transcripts of the session that you heard last week. One thing we were struck by in the evidence you received was the suggestion that there may be miscarriages of justice that can flow from inadequate recording or transcription. As a result of reading that, the three of us considered whether we knew of any instance where there had been a miscarriage of justice on either of those two bases. Our researches have not identified any case where the Court of Appeal has had to consider a potential miscarriage of justice arising from the inadequacy of the recording or transcription.
In so far as the risk is concerned, you probed the witnesses you heard from last week on the basis that, if there is just a transcript provided, for example, of a defendant’s interview, as opposed to hearing the recording itself, there may be something lost in it. Our experience in practice is that there is not. Because we have the ability to check, by way of listening to an interview, we also therefore have the ability to understand whether there is that nuance that needs to be placed before a jury. We do that by reference to identifying what the issues in the case are.
When we know what the issues in the case are, our job is to identify the evidence that will help a jury resolve those issues. If there is a passage in an interview that relates to a particular topic about which something is said in a particular way, and the parties feel that that is relevant to the issues to be resolved, the parties can choose to play that excerpt for a jury’s consideration, but we do not always need to do that.
One other thing we must focus on is why it is that transcripts are used in the way that they are. The time that we have to conduct criminal litigation is limited. If we played a recording of a defendant’s interview in a murder trial, for example, we would lose three to four days. I say “lose”; I use that term advisedly, because that is not time that we could spend elsewhere. A murder suspect will be interviewed for several hours over several days. Much of what is said in the interview is not material that needs to be placed before a jury.
For example, at the outset of each interview a suspect is cautioned. A great deal of time is then taken to explain the caution to the suspect and the suspect is required to confirm that they understand it. We do not need pages of that before the jury. More importantly, we do not need to play that as a recorded part of the evidence before the jury. Instead, we summarise that part to say, “The suspect was cautioned. He or she was told, ‘You do not have to say’, et cetera, and the suspect confirmed that they understood the caution”. Within a few lines, we have summarised what would otherwise take a substantial period in court.
We then get to the issue of repetition. If material is repeated in an interview, that repetition will not assist anyone, so we can remove that. We can certainly remove it from a transcript before it is placed before the jury, or, if there is repetition, there can be a summary: “The issues dealt with at such and such a time were dealt with again and the answers were repeated”. We cannot do that with a recording.
A recording of a suspect’s interview can be edited, but it is a labour-intensive exercise. To put it bluntly, one might ask rhetorically, “Do the police have the resources and time to devote to that process of editing in circumstances where it may not in fact be necessary?” Where it is necessary, it can be done. I have certainly been involved in trials where we have played a suspect’s interview because it was relevant to the issues.
Baroness Cass: I have a very quick technical question. Plainly, there is investment in AI transcription, which is coming later. In terms of the investment to do that, there would presumably be a relatively small investment to switch to digital recording, which would be more readily editable.
Harpreet Sandhu: It is digitally recorded now.
Baroness Cass: Okay. Is there software that would allow you to slice to a particular bit relatively quickly?
Harpreet Sandhu: There may well be. In theory, of course that should be right, but the theory and what happens in practice do not always align in the way that we would anticipate they might.
Fiona Rutherford: I was going to add one other thing. It is very much for the practitioners. I have practised, but both my colleagues beside me are far more recent. In relation to the efficiency point, we know that Sir Brian Leveson has spent a great deal of time engaging with many people and produced a recent report. Maybe we will say something a bit more about potential recommendations this committee can make in relation to his further work.
In terms of efficiency within a trial process, it is a really dynamic situation over the course of days or weeks that you are in court. Taking on board the point that my colleagues have made around the lack of recording and other facilities at court, if things in the trial happen, such as a charge being dropped, which means that big tranches of this interview that has taken place need to be removed from the jury, that can happen very quickly in relation to redacting a transcript. There are other, much more important things in the trial than purely efficiency, but that is just to make that point.
Equally, if a witness does not turn up, which often might happen, the evidence that a defendant or witness in an achieving best evidence—ABE—interview might be asked about would be irrelevant to the interview that is conducted by the police. Again, it should not be put before the jury. I just add those two points.
Q16 The Chair: May I come in on a supplementary? The case you described was, I imagined, a murder trial or a serious crime. The case was taking place over several weeks and you had preparation time, so that all made sense. I know also that sometimes you will not have that preparation time. You will have a case that you have to do tomorrow. That process you described cannot then happen. Do you just go straight to the summary and not listen to the tape?
Harpreet Sandhu: No, you work through the night.
Sam Skinner: The risk is relying on a summary. Summaries are unreliable, not deliberately or because people are skewing matters but because the process of turning a vast amount of evidence into a précis is difficult, and to make it neutral is even more difficult. That will be a risk with transcripts with AI, with whatever biases are factored into the programming. We do not rely on summaries—we never do. We always go to the word-for-word transcript.
The Chair: You would not listen to the tape at that point as you would when you were preparing for a trial that you had advance notice of.
Sam Skinner: We would.
The Chair: You would still listen to the tape.
Sam Skinner: Certainly Harpreet and I would listen to it. We call it a tape. I remember when we used to get the tapes.
The Chair: That is my age. Sorry about that.
Sam Skinner: No, not at all.
The Chair: I mean the recording.
Sam Skinner: I remember people cutting the tapes up to take bits out and playing the tapes. One can do that electronically. One works through the night to check before one stands up in front of a jury in the Crown Court.
The Chair: You have listened to the audio.
Fiona Rutherford: I think that the magistrates’ court is the one place where that might be, in fairness, different, because of the volume.
The Chair: Yes, and because of the nature of the job.
Fiona Rutherford: If you are prosecuting, you might get 15 cases that you might have to prosecute. You know that only three will get on, but you have to have some sense of what each of those cases is about. To apply the diligence, even if you worked every minute that was available, would not be practicable.
The Chair: That is really helpful. Lord Laming, your question to some extent has been asked, but do you want to come in anyway?
Q17 Lord Laming: This is a complex matter for lay people to understand. To use the wrong word, it seems rather haphazard to a lay person, but I am sure that it is not. Could you be clear with us? Who is currently responsible for deciding whether the evidence is presented as a transcript or a recording, and who is responsible for making sure that the evidence is correct and the editing has been handled in an appropriate way?
Harpreet Sandhu: It is counsel.
Sam Skinner: It is the barristers.
Harpreet Sandhu: The barristers are responsible for identifying all evidence that is relevant to the issues in a case. They are responsible for deciding what should be placed before a jury and how it should be placed before a jury.
Sam Skinner: That is also overseen by the judge.
Lord Laming: Can you just explain that to us? Sorry about this. Does this mean that the prosecution and the defence, or whatever it is in other court settings, have to listen to hours and hours of evidence to decide how they want it presented in court?
Sam Skinner: Yes, we do. We listen to it and mark it up against the transcript. We weigh it against the other evidence in the case that is written and decide how best we are going to deploy that to present the case, and equally so when defending. When one defends, it is sometimes the case that one wants to put the recording or transcript of the suspect’s interview with the police before the jury to show demeanour and how he or she reacted when first being taxed with the allegation. We need to have a full, global idea of the evidence that is potentially relevant so we can decide how to present the case, whether prosecuting or defending. It takes hours.
Lord Laming: If it is impossible for the prosecution and the defence to reach agreement, who decides then?
Sam Skinner: The judge.
Lord Laming: This is my last question, sorry.
The Chair: No, please, this is helpful.
Lord Laming: When you say the judge, does that mean that the arguments for the prosecution and the defence are put in public to the judge as part of the trial?
Sam Skinner: Yes. There are certain circumstances where that would not be in public but, 99 times out of 100, it would be in public. Let me give an example where it would not be in public. Where the argument between the parties was about a complainant’s previous sexual history in a rape case, that would be heard by the judge sitting in chambers in private. That sort of material would not be deployed in a public hearing.
Lord Laming: How often does this dispute between the two sets of barristers happen? Is this a once-in-a-lifetime experience or an every-week experience?
Sam Skinner: It is every week. We tend at the Bar to be able to narrow the issue so that the argument is over a single point or a couple of points. It is common for those arguments to take place. It is rare for there to be fully blown arguments about every piece of evidence going.
Q18 The Chair: Ms Rutherford, you might want to take this one. Can I go back a bit? What is the point of the summary if people are not relying on it?
Fiona Rutherford: In the magistrates’ court it is worth emphasising that there is, necessarily, a different approach in relation to using the material that you have. If there is a guilty plea being offered by the defendant, as a prosecutor you would have to weigh up whether that is the right thing to do. You have to work on the basis of the amount of information you have in front of you as well as making phone calls, et cetera, at court if you need to ask further questions.
From a defence perspective, you are likely to have fewer cases in a magistrates’ court than the prosecution would. Your ability to take instructions from your client, which is another really important part of this process, is greater, to get a sense of whether, in a “no comment” interview as the summary describes, it really was a “no comment” or whether there were some comments made and so on.
Q19 Lord Blencathra: What is your impression of the percentage of cases where transcripts only are used and the percentage of cases where the video or audio recording is used in addition to or instead of?
Harpreet Sandhu: Is that a reference to a suspect’s interview under caution?
Lord Blencathra: Generally, are there any court cases where you would use transcripts only?
Harpreet Sandhu: The reason I ask is that we sought to distinguish two potential types of transcript that we use. There is the ABE interview of a witness, for example, who is vulnerable. In those cases, there will always be a transcript and it may be necessary for that transcript to be placed before the jury and then taken away once the recording is heard. In those cases—Sam will correct me if I am wrong—it is routinely done. There will be a recording played with a transcript to assist, if necessary.
Lord Blencathra: I assume that not all cases involve vulnerable witnesses. That must be a tiny minority.
Harpreet Sandhu: It is not necessarily a tiny minority, but you are right to say that they do not exist in all cases. Then we move on to where we have recordings and transcripts of a defendant’s interviews under caution. In those cases, my own experience is that we use transcripts more frequently and there may be occasions when we use some recordings.
Lord Blencathra: Is it 5% or 10%?
Harpreet Sandhu: It is impossible for me to give a figure because it would be inaccurate and therefore unhelpful. Each case has to be assessed on its own merits. Its own basis will provide the foundation for a decision as to whether any part of a recording is played. There is no hard and fast rule about it.
Q20 Baroness Pidgeon: At the start, you talked about the court estate and the quality of the playback equipment being insufficient. Does that have an impact? I guess that you know the courts and the circuit that you work on. Does that have an impact on whether you would choose to play back because you know that, in a certain court, it is not going to help the case, or do you always insist, even though it may well be very poor quality?
Harpreet Sandhu: Our insistence may be futile. To give you an example, I know of some courtrooms where the acoustic is such that, the moment you try to play something, it sounds as though we are all sitting in the most enormous goldfish bowl and you cannot hear a thing. While I know that this is not necessarily an opportunity for us to talk about the court estate generally, your inquiry, although limited in scope, is an example of an issue that is identified but that is symptomatic of a wider issue.
You know what it is. I am afraid to say that successive Administrations have not cared for the criminal justice system. It has been undervalued and underresourced and is now at breaking point. When we think about the ways in which we could seek to improve an aspect of it, it becomes difficult, if not artificial, to speak about how we might, for example, improve the recording or playback equipment in a courtroom, but at the same time not contend with the wider aspects. Even if we have great recording equipment, what do we do about the roof that continues to leak into several buckets in a corner of a room? That might interfere with somebody’s ability to concentrate on the recording, even if it does not necessarily distract from their ability to hear the recording with absolute clarity.
Sam Skinner: May I give a practical example? Say that there is a prosecution for rape and the complainant has given his or her account on the video. The video quality is not good; it is difficult to hear what the witness is saying. When we play it in court to the jury, in a courtroom this size with a screen on this wall and the jury sitting in that corner over there, looking across a large room at a very small screen and a badly focused camera, we cannot hear what the witness is saying.
I have to give the jury a copy of the transcript so they can follow along with what the witness is saying, but they cannot read the transcript at the same time as watching the witness, so the effect of the witness’s demeanour and the way that they are explaining what has happened to them and been done to them is lost, because the jury members have their heads in the transcript and are turning the page.
We therefore invite the police to put subtitles or surtitles on to the screen itself, but then we are into a further degree of complexity. Again, we still lose the ability for the jurors properly to assess how a witness gives their evidence or how a complainant tells their story because of the low quality of the recording and the playback equipment.
Q21 Lord Laming: Can you say a little more about the way in which the system has deteriorated to the point where you cannot achieve what you want to achieve? In government, there is the Lord Chancellor who is responsible for the court system. We now have a Prime Minister who is a distinguished lawyer. Why has the message not got across to government about how serious these issues are that you are raising? I find it really rather disturbing that you have a court system that does not even have the basics of modern technology working satisfactorily.
Sam Skinner: It is not for lack of trying, I can tell you that. The message has not got across, but that is not for lack of trying on behalf of the profession.
Fiona Rutherford: For continuous years, the MoJ has been a non-ring-fenced department in relation to funding. Year after year, it has reduced its funding. It has a huge amount of remit in terms of what its reduced funding has to cover, including prisons, probation and legal aid, for example. Until we all start to be able to better land the value of the justice system—it is heartening to hear your perspective on this—we will continue to struggle to find the funding that is absolutely essential.
Harpreet Sandhu: It is the cornerstone of our democracy.
Lord Laming: For those of us who believe that justice is one of the cornerstones of our society, what you are saying to us is really quite disturbing.
Harpreet Sandhu: Can I give you an example? Every five years, we will hear somebody say, “If you were to elect us, we would provide justice for victims”. I pick those three words, “justice for victims”, advisedly, because they never, it seems to me at least, actually think about how we turn complainants into victims. You do not become a victim in the legal sense as soon as you say, “I am the victim of a crime”. You are a proven victim once we are able to prove that the allegation you advance is an allegation that is right.
How do we do that? By making sure that the trial process is effective. How do we make the trial process effective? By making sure that we have courtrooms that are capable of hearing the cases and a sufficient number of lawyers around the country who are able to prosecute and defend. Incidentally, we do not, because, rather than seeing an increase, we have seen a decrease.
The Chair: We are going a tiny bit off track here, although it is both important and interesting. Lord Carter, were you going to come in there?
Q22 Lord Carter of Coles: Perhaps this is off track. Clearly, you have identified many failures in the system. Why has the Court of Appeal not then overturned serial judgments? If it is not working, I thought that the Court of Appeal was the place that dealt with it.
Harpreet Sandhu: We are not saying that it is not working. Your inquiry is about what needs improving.
Lord Carter of Coles: No, I think you were making the point that it was not working. I want to get the record clear.
Harpreet Sandhu: It works because those who are engaged in it do all that they can to make sure that the system continues to work.
Fiona Rutherford: Also, it sometimes does not work on the basis that we know that trials are set three or four years in advance. There are multiple cases that are not being heard, which means that victims do not get justice, and defendants equally do not.
Q23 The Chair: I am trying to pick up some pieces of background briefing that we have had or things we have heard that we have not covered. There is this phrase, “acting out”. We know what it means, but it has not necessarily been spoken about a lot in what you have said so far. Does that happen a lot and what role does that play?
Harpreet Sandhu: We invite you to approach that with some degree of care. When that term is used, it suggests that we enter into the realms of amateur dramatics, that one person adopts the role of a defendant, another of a police officer, and that there is some degree of unfairness in the way in which things are presented. Our experience is that that does not happen, because those who are tasked with presenting the evidence undertake the task responsibly. Therefore, in our view and experience, there is not this wholesale hamming up, for want of a better phrase, of what is said and how it is said. Does that help on that issue?
Q24 Baroness Pidgeon: In terms of the importance of getting the transcript in front of the jury, do you have problems in terms of the administration of the court and having the copies ready for your jury? Are there delays there in the system as well?
Harpreet Sandhu: There can be. For example, if we are reaching the stage where it is time to present a defendant’s interviews, to take Lady Morris’s example of somebody who receives a case very late, and both people receive a case late, they are working on that interview as other evidence is being heard, often after court, early in the morning, et cetera. They will reach a stage where we are about to present the interviews, because they are in their final edited form, and somebody will find that the photocopiers have broken in the courtroom, to give a homely example. That is an example of the way in which the court infrastructure does not necessarily help. That is an example of how things can break down, which of course results in delay.
Baroness Pidgeon: Therefore you cannot sit because that is the next bit of the trial.
Harpreet Sandhu: You cannot sit when you want to, at least.
Baroness Pidgeon: Yes. I had picked that up elsewhere, that they could not get stuff photocopied, and therefore delay, delay, delay.
Q25 The Chair: Can we go from leaking roofs and poor audio‑visual equipment to future technology? I know that it is a bit of a jump. I appreciate that you have to put the basic infrastructure in courts in place, but we talk as a society about the potential of AI and what computers can do. We have heard some evidence already about this, so I wondered whether we might have a short discussion on that as to what the potential is. Mr Skinner, do you want to start on this one? AI is a bit of a global term at the moment. To ask the broad question, in terms of development of modern technology, could that change things materially in this area? Could the future of AI change it even further?
Sam Skinner: Yes, and nobody is against the use of AI to make things more efficient and cheaper. We get access to justice by doing that. Where we are sceptical is the lack of any evidence base for AI’s ability properly to transcribe. We are sceptical that this will be done without any proper trialling process around the country to see whether AI can be relied upon. We are sceptical about AI’s ability neutrally to summarise anybody’s account, whether it is a suspect, later a defendant, or a complainant or witness, later a victim.
So far as AI’s ability already within the courts to check transcripts, by which I mean word-for-word transcripts, that is something that is being used already. There is a further check, which is us, because we will always double-check what the computer has told us. The short answer is that it is exciting. It could be very useful, but there may be a temptation to see it as a solution to a problem that may or may not exist and to put outsized focus or attention on AI when the roof is leaking in courtroom number one.
The Chair: Yes, I appreciate that.
Fiona Rutherford: Justice has done a fair bit of work in relation to looking at how AI could be implemented more generally. I will say a couple of sentences around the general context. The need for any kind of AI or technology to be installed has to be on a principled or goal-led basis. First, you need to ensure that it is trustworthy when it is being designed and developed. Secondly, you need to ensure that it is delivering a positive impact and value for money. Thirdly, you need to ensure, we would say, that it upholds the rights and the rule of law, importantly, particularly in a high-stakes context such as criminal justice, where people are potentially losing their liberty or otherwise.
I just flag our report, which sets out a rights-based approach, draws on concrete and well-understood enforceable legal rights and provides two clear requirements that may be of assistance. The first is that, when technology or AI is being designed, you have to ensure that the tool is clearly aimed at improving one or more of the justice system’s core goals of access to justice.
The second is a duty to act responsibly. We know that lots of the technology being produced is not being produced by public services or government; it is being produced by private actors, which is fine, but that duty to act responsibly should be at the very heart of what is being designed and delivered in the criminal justice context.
To then apply it more specifically to what this committee is looking at in relation to transcripts, we think that there are potential opportunities in relation to efficiency and timeliness. First, specifically, automated transcription could—I emphasise “could”—reduce delays and costs, particularly in those courts that we have just talked about where we are facing significant backlogs. Faster access to transcripts could support appeals. It could also improve the procedural fairness.
Second is resource allocation. The police and legal professionals, as we have heard, face significant resource constraints. Automated transcription could handle routine tasks, allowing staff to focus on complex analysis instead. Automated tools can also assist in quality assurance, identifying inconsistencies or even omissions in the human-produced transcripts.
I mentioned equal access to justice, which is obviously at the heart of the system itself. Translation tools, powered by automated systems, could help non-native speakers to understand and participate in proceedings. Of course, technology can support individuals who have to represent themselves, of whom there are many across the criminal justice system, helping them to navigate the system more effectively.
Finally, there is scrutiny and public engagement. This is pretty much at the heart of it, because automated transcription could enable comprehensive recordings of court proceedings. It could enable greater open justice and greater understanding of the justice system. To your point, Lord Laming, the ability to raise awareness of the value of the system could be enhanced further, but it has to be done, we would say, with those original goals in mind, and to have a risk-assessed approach as well.
Lord Blencathra: Would it not be better just to have good video and audio recording you can play on every occasion, rather than depend on someone transcribing it, either 320 people in the police service or AI?
Harpreet Sandhu: The transcripts will always be necessary for case preparation purposes. We always need them.
Q26 The Chair: I know the two of you are in the Midlands, but you might work with different police forces. Our next panel is police officers. There is not a national framework for who transcribes and the quality. Do you notice a difference? Do you think to yourself, “We’re in so-and-so’s police area next. The transcription will be slightly different, better or worse?” or are you not aware of any difference?
Sam Skinner: I cannot say I have noticed a difference. I work in London, the south-east and the Midlands. I have worked in the north as well. I cannot say, hand on heart, that I can give evidence that there is a problem with the accuracy of transcriptions for suspects’ interviews that I have seen in the last 18 years.
The Chair: So no one has noticed a difference. I will go on to the last question. We will produce a report, probably quite a short report, but we may want to put in some recommendations to government. We can only recommend to government, by the way. Our reports are to government. Other than spending lots more money on the fabric of the court buildings, which we might well include anyway, is there anything specifically that you think would be a good idea for us to include as a recommendation?
Fiona Rutherford: I mentioned Leveson before. We have part 2 of his report, which he will be starting imminently, we understand. There could be a recommendation that recognises transcription as a key part of the process. Achieving efficiency there could be one of the things that would be appropriate for the review.
The Chair: You are saying efficiency, not accuracy.
Fiona Rutherford: I think it is the two, actually. They are almost the same. It is two sides of the same coin. If there is any potential technology out there, we would recommend piloting before trying. That is an obvious one. Equally, on the point you made around the disparate police forces, you have talked about the 43 in England and Wales. Of course, cases sometimes cross borders. We have Police Scotland and Northern Ireland police as well. If there were a standardisation or a national framework for police transcription, so that you could embed standards across forces, that would be an incredibly useful recommendation. My other recommendations include investment, but in training and equipment—training the people who are transcribing and/or quality assuring—and of course oversight of any AI that is put in place.
The Chair: It is not that we want to ignore it. It is just that that will definitely be there. We can say that with some certainty. Mr Sandhu and Mr Skinner, do you have any recommendations for us?
Harpreet Sandhu: I have nothing to add.
Sam Skinner: The system will work if the people within the system are motivated, not demoralised and properly paid. It is the people who prevent miscarriages of justice, not the computers.
The Chair: Thank you very much. That has been very informative and interesting. Certainly, my own knowledge of the system and the mechanics of it, how it actually works, rather than the theory, has been greatly increased. We are very grateful for your time today and we will now close this session.
